A safe, reliable and multi-hop ad-hoc wireless communication method in a substation

Abstract

The present invention relates to the field of wireless communication technologies, and discloses a safe, reliable and multi-hop ad-hoc wireless communication method in a substation, including the following steps: Step S101, determining the neighbor set of each node according to the communication range of the nodes in the substation; Step S104, generating a search range with the Euclidean distance between the starting node and the ending node as the diameter, and constructing graph network data according to the nodes within the search range; Step S105, inputting the graph network data into a neural network model to output a first adjacency matrix; Step S106, sending the first adjacency matrix to the Kth hopping node to complete communication; The present invention analyzes the association relationship of the nodes in the substation through a neural network model, and generates training samples for training the neural network model through an ant colony algorithm according to the node communication situation at a future time, so that the neural network model has the ability to predict the node communication situation at a future time, thereby improving the fault tolerance rate of wireless communication.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and more specifically, to a safe and reliable multi-hop ad-hoc wireless communication method within a substation. Background Art

[0002] With the continuous improvement of the automation level of the power system, the devices within the substation need to achieve efficient and reliable communication connections. The traditional multi-hop ad-hoc wireless communication method usually sets the jump nodes according to the communication range of the nodes for node-by-node communication. However, with the increase in the total number of transmission jump nodes, the end-to-end transmission delay will show a significant increasing trend. Especially in the case of large-scale concurrent communication, the traditional multi-hop ad-hoc wireless communication method lacks an effective resource scheduling mechanism, and there may be idle nodes, resulting in low node utilization.

[0003] The existing multi-hop ad-hoc wireless communication method selects the jump nodes by calculating the communication quality between nodes (such as channel fading factor, signal-to-noise ratio, etc.), so as to reduce the total number of transmission jump nodes and reduce the end-to-end transmission delay. However, the existing multi-hop ad-hoc wireless communication method mainly relies on the current node communication situation and lacks the ability to predict the future node communication situation. When the node communication situation changes suddenly (such as traffic surges, node failures, etc.), it is difficult to quickly adjust the transmission path, thus affecting the reliability and efficiency of communication. Summary of the Invention

[0004] The present invention provides a safe and reliable multi-hop ad-hoc wireless communication method within a substation to solve the technical problems in the above background art.

[0005] The present invention provides a safe and reliable multi-hop ad-hoc wireless communication method within a substation, including the following steps:

[0006] Step S101, determining the neighbor set of each node according to the communication range of the nodes within the substation;

[0007] Step S102, calculating the signal-to-noise ratio of each node from the starting node to its neighbor set, and selecting the node with the largest signal-to-noise ratio as the first jump node;

[0008] Step S103, calculating the signal-to-noise ratio of each node from the first jump node to its neighbor set, and selecting the node with the largest signal-to-noise ratio as the second jump node, and repeating this step until the Kth jump node is obtained;

[0009] where K is a user-defined parameter;

[0010] Step S104, generating a search range with the Euclidean distance between the starting node and the ending node as the diameter, and constructing graph network data according to the nodes within the search range;

[0011] The graph network data G is represented as: G(V, E), where V represents the set of vertices in the graph network data G, and E represents the set of edges between the vertices in the graph network data G;

[0012] The i-th vertex establishes a mapping relationship with the i-th node, where 1 ≤ i ≤ N, and N represents the number of nodes within the search range;

[0013] The initial feature of the vertex is represented by the attribute of the node with which it establishes a mapping relationship;

[0014] Step S105, input the graph network data into the neural network model and output the first adjacency matrix;

[0015] The first adjacency matrix includes N rows and N columns. The element values of the first adjacency matrix are represented by 0 or 1, that is, the element value of the m-th row and the n-th column being 1 indicates that the m-th node jumps to the n-th node, otherwise it indicates that the m-th node does not jump to the n-th node, where 1 ≤ m ≤ N, 1 ≤ n ≤ N;

[0016] Step S106, send the first adjacency matrix to the K-th jump node to complete communication.

[0017] Furthermore, the signal-to-noise ratio from the i-th node to the j-th node has the following calculation formula:

[0018] ;

[0019] ;

[0020] ;

[0021] ;

[0022] where represents the useful signal power received by the j-th node, represents the noise signal power received by the j-th node, represents the signal transmission power of the i-th node, and respectively represent the channel fading factor and the Euclidean distance from the i-th node to the j-th node. Temp represents the absolute temperature in Kelvin, represents the bandwidth of the j-th node, provided by the device manufacturer in Hertz, where , c represents the speed of light, f represents the carrier frequency, and k represents the Boltzmann constant.

[0023] Furthermore, the attributes of the nodes include: received signal strength, maximum transmission rate, packet loss rate, memory size, and link quality indication, and the initial features of the corresponding vertices are obtained by normalizing the attributes of the nodes.

[0024] Furthermore, calculate the correlation coefficient between each node within the search range and all nodes in its neighbor set, and when the correlation coefficient is greater than or equal to the set threshold, an edge connection is constructed between the corresponding two vertices, where the set threshold is a custom parameter;

[0025] The correlation coefficient between the i-th node and the j-th node is calculated as follows:

[0026] ;

[0027] where Y represents the number of dimensions of the initial features of the vertices, and respectively represent the average values of the initial features of the vertices corresponding to the i-th node and the j-th node, represents the y-th dimension value of the initial features of the vertex corresponding to the i-th node, represents the y-th dimension value of the initial features of the vertex corresponding to the j-th node.

[0028] Furthermore, the neural network model includes: a first hidden layer, a second hidden layer, and a third hidden layer;

[0029] The first hidden layer is used to convert the initial features of each vertex of the graph network data into a matrix representation;

[0030] The calculation formula of the first hidden layer is as follows: ;

[0031] where represents the matrix representation of the i-th vertex, represents the initial features of the i-th vertex, and W represents the weight parameter;

[0032] The second hidden layer extracts features from the matrix representation of each vertex through 3 convolutional kernels of different sizes to obtain 3 feature maps of different sizes, and after expanding the 3 feature maps into vector representations and splicing them, a combined vector is obtained, where the sizes of the 3 convolutional kernels are 1×1, 3×3, and 5×5 respectively, and the stride of the 3 convolutional kernels is 1;

[0033] The third hidden layer inputs the graph network data and outputs the first adjacency matrix.

[0034] Furthermore, the calculation formula of the third hidden layer includes:

[0035] ;

[0036] ;

[0037] ;

[0038] Where S represents the first adjacency matrix output by the third hidden layer, represents the set of vertices that have an edge connection with the i-th vertex, represents the number of elements in, and respectively represent the combined vectors of the i-th vertex and the j-th vertex, represents the updated vector of the i-th vertex, represents the correlation coefficient between the i-th vertex and the j-th vertex, and respectively represent the first weight parameter and the second weight parameter, and MLP represents a multi-layer perceptron, represents stacking the updated vectors of N vertices, concat represents the concatenation operation, T represents the transpose operation, sigmoid represents the sigmoid activation function, and F represents rounding up the element values greater than or equal to 0.5 to 1 and rounding down the element values less than 0.5 to 0.

[0039] Furthermore, sample data for training the neural network model is obtained according to steps S101 to S104. After a preset time period, the attributes of all nodes in the substation are collected again. A transmission path is generated through the ant colony algorithm and converted into a second adjacency matrix representation as the sample label for the training sample of the neural network model. The transmission path is represented by a vector with a dimension of N, and the element values of the vector are represented by 0 or 1. The i-th dimension value of 1 indicates passing through the i-th node, otherwise it indicates not passing through the i-th node, where the preset time period is a custom parameter.

[0040] Furthermore, the difference between the first adjacency matrix output by the neural network model at each iteration and the second adjacency matrix is specified as the loss function until the maximum number of iterations is reached, where the maximum number of iterations is a custom parameter.

[0041] The present invention provides a safe and reliable multi-hop ad hoc wireless communication system in a substation, including:

[0042] A neighbor set determination module for determining the neighbor set of each node according to the communication range of the nodes in the substation;

[0043] A first jump module for calculating the signal-to-noise ratio of the starting node to each node in its neighbor set and selecting the node with the maximum signal-to-noise ratio as the first jump node;

[0044] A second jump module, which is used to calculate the signal-to-noise ratio between the first jump node and each node in its neighbor set, and select the node with the largest signal-to-noise ratio as the second jump node, and repeat this step until the Kth jump node is obtained;

[0045] A graph network data construction module, which is used to generate a search range with the Euclidean distance between the starting node and the ending node as the diameter, and construct graph network data based on the nodes within the search range;

[0046] A first adjacency matrix generation module, which is used to input the graph network data into a neural network model and output a first adjacency matrix;

[0047] A first adjacency matrix sending module, which is used to send the first adjacency matrix to the Kth jump node to complete communication.

[0048] The present invention provides a readable storage medium, which stores non-transitory computer-readable instructions. When the non-transitory computer-readable instructions are executed by a computer, the steps of the above-mentioned secure and reliable multi-hop ad hoc wireless communication method in a substation are executed.

[0049] The beneficial effects of the present invention are as follows: The present invention analyzes the association relationship between nodes in a substation through a neural network model, and generates training samples for training the neural network model through an ant colony algorithm according to the node communication situation in the future time, so that the neural network model has the ability to predict the node communication situation in the future time, thereby improving the fault tolerance rate of wireless communication. Brief Description of the Drawings

[0050] Figure 1 is a flowchart of a secure and reliable multi-hop ad hoc wireless communication method in a substation according to the present invention;

[0051] Figure 2 is a schematic diagram of a secure and reliable multi-hop ad hoc wireless communication system in a substation according to the present invention;

[0052] Figure 3 is a comparison diagram of experimental results of the present invention.

[0053] In the figure: neighbor set determination module 201, first jump module 202, second jump module 203, graph network data construction module 204, first adjacency matrix generation module 205, first adjacency matrix sending module 206. Detailed Embodiments

[0054] Reference will now be made to example embodiments to discuss the subject matter described herein. It should be understood that discussing these embodiments is only to enable those skilled in the art to better understand and thus implement the subject matter described herein. Without departing from the scope of protection of the content of this specification, changes can be made to the functions and arrangements of the elements discussed. Each example can omit, substitute, or add various processes or components as needed. Additionally, the features described relative to some examples can also be combined in other examples.

[0055] It should be noted that unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should have the ordinary meaning understood by those of ordinary skill in the art to which the present invention pertains. The terms "first", "second", and similar words used in one or more embodiments of the present invention do not denote any order, quantity, or importance, but are only used to distinguish different components. Words such as "including" or "comprising" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. Words such as "connected" or "coupled" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Upper", "lower", "left", "right", etc. are only used to represent relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0056] As Figures 1 to 3 shown, a safe and reliable multi-hop ad-hoc wireless communication method in a substation includes the following steps:

[0057] Step S101, determine the neighbor set of each node according to the communication range of the nodes in the substation;

[0058] Step S102, calculate the signal-to-noise ratio of the starting node to each node in its neighbor set, and select the node with the maximum signal-to-noise ratio as the first hop node;

[0059] Step S103, calculate the signal-to-noise ratio of the first hop node to each node in its neighbor set, and select the node with the maximum signal-to-noise ratio as the second hop node, and repeat this step until the Kth hop node is obtained;

[0060] where K is a user-defined parameter. Preferably, K is set to 3;

[0061] Step S104, generate a search range with the Euclidean distance between the starting node and the terminating node as the diameter, and construct graph network data based on the nodes within the search range;

[0062] The graph network data G is represented as: G(V, E), where V represents the set of vertices in the graph network data G, and E represents the set of edges between the vertices in the graph network data G;

[0063] The i-th vertex establishes a mapping relationship with the i-th node, where 1 ≤ i ≤ N, and N represents the number of nodes within the search range;

[0064] The initial feature of the vertex is represented by the attribute of the node with which it establishes a mapping relationship;

[0065] Step S105: Input the graph network data into the neural network model and output the first adjacency matrix;

[0066] The first adjacency matrix includes N rows and N columns. The element values of the first adjacency matrix are represented by 0 or 1. That is, if the element value in the m-th row and n-th column is 1, it means the m-th node jumps to the n-th node; otherwise, it means the m-th node does not jump to the n-th node, where 1 ≤ m ≤ N and 1 ≤ n ≤ N;

[0067] Step S106: Send the first adjacency matrix to the K-th jump node to complete the communication.

[0068] It should be noted that selecting the jump node according to the signal-to-noise ratio between nodes can directly perform data transmission before generating the first adjacency matrix, avoiding the situation where the time for generating the first adjacency matrix is too long, resulting in data transmission blockage and inability to transmit normally. Therefore, this "asynchronous" design can effectively improve the overall communication efficiency.

[0069] It should be noted that the nodes in the substation represent communication devices installed in the substation or intelligent devices with communication functions, such as sensors, intelligent circuit breakers, protection devices, etc. Each node in the substation is assigned a unique code, such as an incrementing positive integer code or UUID, etc. The first adjacency matrix represents the two-dimensional matrix encoding of the transmission path from the starting node to the ending node. For example, if the element value in the 3rd row and 5th column of the first adjacency matrix is 1, it means the 3rd node jumps to the 5th node.

[0070] In an embodiment of the present invention, the signal-to-noise ratio from the i-th node to the j-th node The calculation formula includes:

[0071] ;

[0072] ;

[0073] ;

[0074] ;

[0075] where Indicates the useful signal power received by the j-th node, Indicates the noise signal power received by the j-th node, Indicates the signal transmission power of the i-th node, and respectively represent the channel fading factor and the Euclidean distance from the i-th node to the j-th node. Temp represents the absolute temperature in Kelvin, Indicates the bandwidth of the j-th node, provided by the device manufacturer in Hertz, where , c represents the speed of light, f represents the carrier frequency, and k represents the Boltzmann constant.

[0076] In an embodiment of the present invention, the attributes of a node include: received signal strength, maximum transmission rate, packet loss rate, memory size, and link quality indicator; the initial features of the corresponding vertices are obtained by normalizing the attributes of the nodes.

[0077] It should be noted that the initial features of the vertices are represented by a 1×5 vector. The normalization process can be Min-Max normalization or Z-score normalization. The Link Quality Indicator (LQI) is usually an index used in wireless communication to evaluate the quality of packet reception, reflecting the reliability of packets during transmission, and is commonly used in wireless sensor networks (such as Zigbee networks) and other low-power wide-area network (LPWAN) technologies. The LQI value is usually an integer between 0 and 255, where 255 represents the best link quality and 0 represents the worst link quality.

[0078] In an embodiment of the present invention, calculate the correlation coefficient between each node within the search range and all nodes in its neighbor set, and when the correlation coefficient is greater than or equal to the set threshold, an edge connection is constructed between the corresponding two vertices, where the set threshold is a custom parameter. Preferably, the set threshold is set to 0.5;

[0079] The correlation coefficient between the i-th node and the j-th node is calculated as follows:

[0080] ;

[0081] where Y is assigned the value of 5, and respectively represent the average values of the initial features of the corresponding vertices of the i-th node and the j-th node, represents the y-th dimension value of the initial feature of the corresponding vertex of the i-th node, represents the y-th dimension value of the initial feature of the corresponding vertex of the j-th node.

[0082] In one embodiment of the present invention, the neural network model includes: a first hidden layer, a second hidden layer, and a third hidden layer;

[0083] The first hidden layer is used to convert the initial features of each vertex of the graph network data into a matrix representation;

[0084] The calculation formula of the first hidden layer is as follows: ;

[0085] where represents the matrix representation of the i-th vertex, represents the initial feature of the i-th vertex, and W represents the weight parameter;

[0086] For example, W is designed as a vector of size 32×1, and the initial feature of the vertex is represented by a vector of size 1×5. The resulting matrix size after multiplying the two is 32×5;

[0087] The second hidden layer performs feature extraction on the matrix representation of each vertex through 3 convolution kernels of different sizes to obtain 3 feature maps of different sizes, and unfolds and splices the 3 feature maps into a vector representation to obtain a combined vector. The sizes of the 3 convolution kernels are 1×1, 3×3, and 5×5 respectively, and the stride of the 3 convolution kernels is 1;

[0088] For example, if the matrix size of the vertex is 32×5, the sizes of the 3 feature maps obtained through the 3 convolution kernels are 32×5, 30×3, and 28×1 respectively. Then the size of the combined vector obtained by unfolding and splicing is 1×(160 + 90 + 28) = 1×278;

[0089] The third hidden layer inputs the graph network data and outputs the first adjacency matrix.

[0090] It should be noted that the second hidden layer may include multiple convolution kernels, and after performing convolution operations, feature extraction may also be performed through max pooling or average pooling, which will not be elaborated here.

[0091] In one embodiment of the present invention, the calculation formula of the third hidden layer includes:

[0092] ;

[0093] ;

[0094] ;

[0095] where S represents the first adjacency matrix output by the third hidden layer, represents the set of vertices that have an edge connection with the i-th vertex, represents the number of elements in and respectively represent the combined vector of the i-th vertex and the j-th vertex, represents the updated vector of the i-th vertex, represents the correlation coefficient between the i-th vertex and the j-th vertex, and respectively represent the first weight parameter and the second weight parameter, and MLP represents a multi-layer perceptron, represents stacking the updated vectors of N vertices, concat represents the concatenation operation, T represents the transpose operation, sigmoid represents the sigmoid activation function, and F represents rounding up the element values greater than or equal to 0.5 to 1 and rounding down the element values less than 0.5 to 0.

[0096] It should be noted that both the first weight parameter and the second weight parameter in the third hidden layer are learnable hyperparameters. For example, the size of the combined vector of vertices is 1×278, and the first weight parameter and the second weight parameter can be designed as matrices of size 278×8. Then the size of the updated vector is 1×8, The result of

[0097] is N×N, and the sigmoid activation function is used to unify the element values in the matrix of size N×N between 0 and 1. In an embodiment of the present invention, the sample data for training the neural network model is obtained according to steps S101 to S104. After a preset time period, the attributes of all nodes in the substation are collected again. The transmission path is generated by the ant colony algorithm and converted into a second adjacency matrix representation as the sample label for the training sample of the neural network model. The transmission path is represented by a vector with a dimension number of N, and the element values of the vector are represented by 0 or 1. The i-th dimension value being 1 means passing through the i-th node, otherwise it means not passing through the i-th node, where the preset time period is a custom parameter. Preferably, the preset time period is set to 30 seconds.

[0098] In an embodiment of the present invention, the heuristic factor from the i-th node to the j-th node is calculated as follows:

[0099] ;

[0100] where , , and respectively represent the maximum transmission rate, packet loss rate, link quality indication, and received signal strength from the i-th node to the j-th node.

[0101] It should be noted that generating a transmission path through the ant colony algorithm after a preset time period is to enable the neural network model to have the predictive ability for node communication in the future, thereby improving the fault tolerance rate of wireless communication to a certain extent.

[0102] In one embodiment of the present invention, the difference between the first adjacency matrix and the second adjacency matrix output by the neural network model for each iteration number is specified as the loss function until the maximum iteration number is reached, where the maximum iteration number is a custom parameter. Preferably, the maximum iteration number is set to 500.

[0103] It should be noted that the gradient information of the neural network model with respect to each weight parameter is calculated through the chain rule, and the weight parameters in the neural network model are updated through the gradient descent algorithm to minimize the loss function, which will not be elaborated here.

[0104] In one embodiment of the present invention, as Figure 2 shown, the present invention provides a safe and reliable multi-hop ad-hoc wireless communication system in a substation, including:

[0105] A neighbor set determination module 201, which is used to determine the neighbor set of each node according to the communication range of the nodes in the substation;

[0106] A first jump module 202, which is used to calculate the signal-to-noise ratio of the starting node to each node in its neighbor set, and select the node with the largest signal-to-noise ratio as the first jump node;

[0107] A second jump module 203, which is used to calculate the signal-to-noise ratio of the first jump node to each node in its neighbor set, and select the node with the largest signal-to-noise ratio as the second jump node, and repeat this step until the Kth jump node is obtained;

[0108] A graph network data construction module 204, which is used to generate a search range with the Euclidean distance between the starting node and the ending node as the diameter, and construct graph network data according to the nodes within the search range;

[0109] A first adjacency matrix generation module 205, which is used to input the graph network data into the neural network model and output the first adjacency matrix;

[0110] A first adjacency matrix sending module 206, which is used to send the first adjacency matrix to the Kth jump node to complete the communication.

[0111] In one embodiment of the present invention, the present invention provides a readable storage medium, which stores non-temporary computer-readable instructions. When the non-temporary computer-readable instructions are executed by a computer, the steps of the above-mentioned safe and reliable multi-hop ad-hoc wireless communication method in a substation are executed.

[0112] AsFigure 3 As shown, the 10-hop communications are completed respectively through calculating the signal-to-noise ratio between nodes, the neural network model provided by the present invention, and the ant colony algorithm. The Euclidean distance between the starting node and the ending node of the hop communication is 1000 m, the size of the data packet sent in the hop communication is 1500 bytes, and the maximum transmission rate is 100 Mbps. It is found through observation that the time taken for the neural network model provided by the present invention to complete the hop communication is close to the time taken for the ant colony algorithm to complete the hop communication.

[0113] The above has described the embodiments of this example, but this example is not limited to the above specific implementation manners. The above specific implementation manners are merely illustrative rather than restrictive. Under the inspiration of this example, those of ordinary skill in the art can also make many forms, all of which fall within the protection scope of this example.

Claims

1. A safe and reliable multi-hop ad hoc wireless communication method in a substation, characterized in that It includes the following steps: Step S101: Determine the neighbor set of each node according to the communication range of the nodes in the substation; Step S102: Calculate the signal-to-noise ratio of the starting node to each node in its neighbor set, and select the node with the maximum signal-to-noise ratio as the first hop node; Step S103: Calculate the signal-to-noise ratio of the first hop node to each node in its neighbor set, and select the node with the maximum signal-to-noise ratio as the second hop node, and repeat this step until the Kth hop node is obtained; where K is a user-defined parameter; Step S104: Generate a search range with the Euclidean distance between the starting node and the ending node as the diameter, and construct graph network data based on the nodes within the search range; The graph network data G is expressed as: G(V, E), where V represents the set of vertices in the graph network data G, and E represents the set of edges between the vertices in the graph network data G; The ith vertex establishes a mapping relationship with the ith node, where 1 ≤ i ≤ N, and N represents the number of nodes within the search range; The initial feature of the vertex is represented by the attributes of the node with which it establishes a mapping relationship; Step S105: Input the graph network data into the neural network model and output the first adjacency matrix; The first adjacency matrix includes N rows and N columns, and the element values of the first adjacency matrix are represented by 0 or 1, that is, the element value of the mth row and the nth column being 1 means that the mth node jumps to the nth node, otherwise it means that the mth node does not jump to the nth node, where 1 ≤ m ≤ N, 1 ≤ n ≤ N; Step S106: Send the first adjacency matrix to the Kth hop node to complete the communication; Obtain the sample data of the training sample for training the neural network model according to steps S101 to S104. After a preset time period, collect the attributes of all nodes in the substation again, generate a transmission path through the ant colony algorithm, and convert the transmission path into a second adjacency matrix representation as the sample label of the training sample for training the neural network model, where the transmission path is represented by a vector with a dimension number of N, and the element values of the vector are represented by 0 or 1. The value of the ith dimension being 1 means passing through the ith node, otherwise it means not passing through the ith node, where the preset time period is a user-defined parameter; Specify the difference between the first adjacency matrix output by the neural network model at each iteration and the second adjacency matrix as the loss function until the maximum number of iterations is reached, where the maximum number of iterations is a user-defined parameter; The neural network model includes: a first hidden layer, a second hidden layer, and a third hidden layer; The first hidden layer is used to convert the initial feature of each vertex of the graph network data into a matrix representation; The calculation formula of the first hidden layer is as follows: ; where represents the matrix representation of the i-th vertex, represents the initial feature of the i-th vertex, and W represents the weight parameter; The second hidden layer performs feature extraction on the matrix representation of each vertex through three convolution kernels of different sizes to obtain three feature maps of different sizes, and expands the three feature maps into vector representations and then splices them to obtain a combined vector, where the sizes of the three convolution kernels are 1×1, 3×3, and 5×5 respectively, and the strides of the three convolution kernels are all 1; The third hidden layer inputs the graph network data and outputs the first adjacency matrix; The calculation formula of the third hidden layer includes: ； ； ； Where S represents the first adjacency matrix output by the third hidden layer, represents the set of vertices that have an edge connection with the i-th vertex, represents the number of elements in, and represent the combined vectors of the i-th vertex and the j-th vertex respectively, represents the updated vector of the i-th vertex, represents the correlation coefficient between the i-th vertex and the j-th vertex, and represent the first weight parameter and the second weight parameter respectively, MLP represents a multi-layer perceptron, represents stacking the updated vectors of N vertices, concat represents the concatenation operation, T represents the transpose operation, sigmoid represents the sigmoid activation function, and F represents rounding up the element values greater than or equal to 0.5 to 1 and rounding down the element values less than 0.5 to 0.

2. A secure and reliable multi-hop ad-hoc wireless communication method within a substation according to claim 1, characterized in that, The signal-to-noise ratio from the i-th node to the j-th node Calculation formula It includes: ； ； ； ； Among them represents the useful signal power received by the j-th node, represents the noise signal power received by the j-th node, represents the signal transmission power of the i-th node, and respectively represent the channel fading factor and the Euclidean distance from the i-th node to the j-th node. Temp represents the absolute temperature in Kelvin. represents the bandwidth of the j-th node, provided by the device manufacturer in Hertz. Among them , where c represents the speed of light, f represents the carrier frequency, and k represents the Boltzmann constant.

3. A secure and reliable multi-hop ad-hoc wireless communication method within a substation according to claim 1, characterized in that, The attributes of the nodes include: received signal strength, maximum transmission rate, packet loss rate, memory size, and link quality indication, and the initial features of the corresponding vertices are obtained by normalizing the attributes of the nodes.

4. A secure and reliable multi-hop ad-hoc wireless communication method within a substation according to claim 1, characterized in that, Calculate the correlation coefficient between each node within the search range and all nodes in its neighbor set, and if the correlation coefficient is greater than or equal to the set threshold, an edge connection is constructed between the corresponding two vertices, where the set threshold is a custom parameter; The correlation coefficient between the i-th node and the j-th node is calculated as follows: ； where Y represents the number of dimensions of the initial features of the vertices, and respectively represent the average values of the initial features of the vertices corresponding to the i-th node and the j-th node, represents the y-th dimensional value of the initial features of the vertices corresponding to the i-th node, represents the y-th dimensional value of the initial features of the vertices corresponding to the j-th node.

5. A safe and reliable multi-hop ad-hoc wireless communication system within a substation, characterized in that, Execute a secure and reliable multi-hop ad-hoc wireless communication method in a substation as described in any one of claims 1 to 4, including: A neighbor set determination module, which is used to determine the neighbor set of each node according to the communication range of the nodes in the substation; A first jump module, which is used to calculate the signal-to-noise ratio of the starting node to each node in its neighbor set, and select the node with the maximum signal-to-noise ratio as the first jump node; A second jump module, which is used to calculate the signal-to-noise ratio of the first jump node to each node in its neighbor set, and select the node with the maximum signal-to-noise ratio as the second jump node, and repeat this step until the Kth jump node is obtained; A graph network data construction module, which is used to generate a search range with the Euclidean distance between the starting node and the ending node as the diameter, and construct graph network data based on the nodes within the search range; A first adjacency matrix generation module, which is used to input the graph network data into a neural network model and output a first adjacency matrix; A first adjacency matrix sending module, which is used to send the first adjacency matrix to the Kth jump node to complete communication.

6. A readable storage medium, characterized in that, It stores non-temporary computer-readable instructions, and when the non-temporary computer-readable instructions are executed by a computer, it executes a secure and reliable multi-hop ad-hoc wireless communication method in a substation as described in any one of claims 1 to 4.

Images