A method and system for high-quality communication of a wireless communication module in a regional range

Abstract

The application relates to the technical field of communication, and discloses a high-quality communication method and system for a wireless communication module in a region, which extracts key elements and element feature information in an industrial field image, integrates the communication node knowledge graph of a target region, forms an enhanced knowledge graph, uses a graph neural network combined with an attention mechanism to learn the graph structure data of the enhanced knowledge graph, captures deep-level correlations between communication nodes and key elements, obtains communication quality prediction requirements, and then solves a communication resource configuration scheme for each communication node according to the influence mapping relationship between communication quality requirement parameters and different communication resources, so as to perform industrial field communication. Thus, the application can accurately capture the dynamic changes and characteristics of equipment, personnel and materials in an industrial scene, adopt the combination of knowledge graph fusion and a graph neural network, realize the early prediction of communication requirement changes, and guarantee the stability and efficiency of system communication.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a high-quality communication method and system for wireless communication modules within a regional area. Background Technology

[0002] In industrial automation scenarios, wireless communication plays a crucial role in transmitting equipment control commands and transmitting production data back to the manufacturer, requiring extremely high levels of real-time performance, reliability, and stability.

[0003] However, industrial sites present complex situations such as diverse equipment, high personnel mobility, and dynamic material allocation. The communication needs of communication nodes (such as various sensors, control terminals, and gateways) change dynamically with the surrounding environment (personnel distribution, equipment operating status, material location, etc.). Traditional communication resource scheduling methods often rely on preset rules or passive feedback mechanisms based on network status, failing to fully explore the deep-seated relationships between communication nodes and key elements (equipment, personnel, materials, etc.) in the industrial scenario. This results in insufficient timeliness and accuracy of resource scheduling, making it difficult to adapt to the dynamically changing industrial environment, and consequently affecting communication quality and the overall operational efficiency of industrial automation systems.

[0004] Therefore, how to accurately perceive the key elements and characteristics of dynamic changes in industrial automation scenarios, realize the advance optimization and scheduling of communication resources, and ensure the high quality and stability of wireless communication in industrial automation scenarios is a technical problem that urgently needs to be solved. Summary of the Invention

[0005] This invention provides a high-quality communication method and system for wireless communication modules within a regional area, aiming to solve at least one of the above-mentioned technical problems.

[0006] To achieve the above objectives, the present invention provides a high-quality communication method for a wireless communication module within a regional area, the method comprising the following steps:

[0007] Collect industrial site images of the target area, and extract several key elements and element feature information of each key element from the industrial site images;

[0008] Obtain communication node feature information for each communication node configured with a wireless communication module within the target area; construct a communication node knowledge graph for the target area based on the communication node feature information; combine the element feature information with the communication node knowledge graph to generate an enhanced knowledge graph.

[0009] The graph structure data of the enhanced knowledge graph is learned by using a graph neural network with an attention mechanism to capture the deep-level association between communication nodes and key elements, and to obtain the communication quality prediction requirements of each communication node output by the graph neural network; wherein, the communication quality prediction requirements include several dimensions of communication requirement parameters.

[0010] Based on the communication quality demand parameters of several dimensions in the communication quality prediction demand, and according to the influence mapping relationship between the communication quality demand parameters and different communication resources, the communication resource configuration scheme of each communication node is solved.

[0011] According to the communication resource configuration scheme, the communication resource configuration of the wireless communication module of each communication node within the area is performed, and the industrial field communication is realized based on the communication resource configuration results.

[0012] Optionally, the steps of acquiring industrial site images of the target area and extracting several key elements and element feature information of each key element from the industrial site images specifically include:

[0013] It receives industrial field images collected and transmitted by several image acquisition devices deployed in the target area, and uses image semantic segmentation to identify and extract several key elements and element features of each key element in the industrial field images.

[0014] For each identified key element, extract the element feature vector from its element features to generate element feature information for each key element; wherein, the key element includes at least one of equipment element, personnel element, and material element, the element features include element location coordinates, element quantity, and element activity status, and the element feature vector includes feature parameters of several dimensions.

[0015] Optionally, image semantic segmentation is used to identify and extract several key elements and the element features of each key element from the industrial site image. Specifically, this includes:

[0016] The received industrial site images are input into the U-Net architecture deep learning model to perform pixel-level classification of the images and identify at least one key element among equipment elements, personnel elements, and material elements, as well as the element features of each key element.

[0017] The U-Net architecture deep learning model uses a loss function that combines cross-entropy loss, Dice loss, and smoothing loss during training, and its specific expression is as follows:

[0018] ;

[0019] In the formula, This represents the loss function used during training of the U-Net architecture deep learning model. Represents cross-entropy loss, This indicates Dice's loss. Indicates smoothing loss. , These are the weighting coefficients.

[0020] Optionally, the step of acquiring communication node feature information for each communication node configured with a wireless communication module within the target area, and constructing a communication node knowledge graph for the target area based on the communication node feature information, specifically includes:

[0021] The node attribute information of each communication node configured with a wireless communication module in the target area is collected in advance, and the node feature vector of each communication node is generated as the communication node feature information.

[0022] Based on the physical connection association, functional association and data flow association between communication nodes, the edges between communication nodes are constructed, and the edge weights are calculated by combining the communication frequency, communication data volume and communication data dependency.

[0023] A knowledge graph of communication nodes for the target region is constructed based on the set of all communication nodes, the set of all edges, and the set of all edge weights.

[0024] Optionally, the expression for calculating the edge weight is as follows:

[0025] ;

[0026] In the formula, This represents the edge weight between communication nodes i and j. This indicates the communication frequency between communication nodes i and j. This represents the amount of data communicated between communication nodes i and j. This represents the data dependency between communication nodes i and j. , , These are the weighting coefficients.

[0027] Optionally, the step of combining the element feature information with the communication node knowledge graph to generate an enhanced knowledge graph specifically includes:

[0028] The element feature vector corresponding to the element feature information Dimensionality reduction is performed, and the reduced element feature vectors are... Communication node feature vector corresponding to communication node feature information Perform feature concatenation to construct a fused feature vector. And perform normalization processing;

[0029] For each key element, a corresponding element node is generated. Based on the preset matching rules between communication nodes and key elements, communication nodes and element nodes that match the matching rules are automatically identified and matched. Related edges are generated between communication nodes and element nodes that match the matching rules.

[0030] Based on the communication node feature vector of each communication node and the element feature vector of each key element Calculate the cosine similarity and Euclidean distance between each communication node and each key element, and calculate the weights according to the preset weight coefficients as the weights of the associated edges.

[0031] An enhanced knowledge graph of the target region is constructed based on the node set consisting of communication nodes and element nodes, the set of associated edges consisting of all associated edges, and the set of associated edge weights consisting of all associated edge weights.

[0032] Optionally, the graph structure data of the enhanced knowledge graph is learned using a graph neural network incorporating an attention mechanism to capture deep-level relationships between communication nodes and key elements, and the communication quality prediction requirement for each communication node output by the graph neural network is obtained. This step specifically includes:

[0033] An initial multi-layer graph neural network is constructed, using historical communication quality parameters of the target area and key element feature data extracted from industrial field images of the same period as the training set. The training objective is to minimize the prediction error, and the multi-layer graph neural network is trained using the training set.

[0034] The specific expression for the training loss function is as follows:

[0035] ;

[0036] In the formula, To predict communication demand, For actual communication needs, i represents the i-th communication node, and N represents the total number of communication nodes. Represents the vector norm;

[0037] The fused feature vector at the current time As initial node features By utilizing a trained multi-layer graph neural network combined with an attention mechanism, and employing a layer-by-layer node feature propagation and aggregation approach, the features of the nodes in the previous layer are utilized. Deriving the node features of the next layer ;

[0038] The expression derived using a multi-layer graph neural network combined with an attention mechanism is as follows:

[0039] ;

[0040] ;

[0041] In the formula, For attention weights, For attention magnitude, ∥ represents the feature concatenation operation. For activation function, For the neighboring nodes of communication node i in the augmented knowledge graph connected by edges or associated edges, For the first Layer trainable weight matrix, For activation functions;

[0042] Based on the final node features obtained in the last layer, the communication quality requirements of the output communication node at the next time step are predicted by performing linear transformation and activation on the feature vector of each communication node. .

[0043] Optionally, the communication quality requirement parameters in the predicted communication quality requirements include: communication transmission rate, communication transmission delay, and communication packet loss rate; the communication resources include communication bandwidth parameters, communication power parameters, and communication frequency parameters. Based on these parameters, and according to the influence mapping relationship between the parameters and different communication resources, the steps for solving the communication resource configuration scheme for each communication node specifically include:

[0044] Based on the communication quality requirement parameters of several dimensions in the communication quality prediction requirements, and according to the influence mapping relationship between communication transmission rate, communication transmission delay and communication transmission packet loss rate and different communication resources, the communication bandwidth parameter, communication power parameter and communication frequency parameter are calculated sequentially.

[0045] The specific expression for the mapping relationship between communication transmission rate and communication resources is as follows:

[0046] ;

[0047] In the formula, This represents the communication transmission rate of communication node i. This represents the spectral efficiency coefficient. The communication bandwidth parameter represents the communication node i.

[0048] The specific expression for the mapping relationship between communication transmission delay and communication resources is as follows:

[0049] ;

[0050] In the formula, This represents the communication transmission rate of communication node i. This represents the communication transmission delay of communication node i. This represents the amount of data per frame for communication node i. Indicates the power attenuation coefficient. This represents the communication power parameter of communication node i. Indicates the physical layer base delay;

[0051] The expression for the mapping relationship between packet loss rate and communication resources is as follows:

[0052] ;

[0053] In the formula, This represents the communication power parameter of communication node i. This represents the packet loss rate during communication transmission of communication node i. Indicates the base packet loss rate. , These are the fitting coefficients. Here are the communication transmission frequency parameters for communication node i. ) indicates frequency Interference with cleanliness;

[0054] Based on the communication bandwidth, communication power, and communication frequency parameters of each communication node, a communication resource configuration scheme is generated for each communication node.

[0055] Optionally, according to the communication resource configuration scheme, communication resource configuration is performed for the wireless communication module of each communication node within the area. Based on the communication resource configuration results, industrial field communication steps are implemented, specifically including:

[0056] Extract the communication bandwidth parameters, communication power parameters, and communication frequency parameters of each communication node recorded in the communication resource configuration scheme;

[0057] The communication bandwidth, communication power, and communication frequency parameters of each communication node are sent to the wireless communication module of the corresponding communication node, driving each communication node to configure the communication configuration parameters of the wireless communication module, and realizing industrial field communication based on the configuration results.

[0058] Furthermore, to achieve the above objectives, the present invention also provides a high-quality communication system for a wireless communication module within a regional area, comprising:

[0059] The acquisition module is used to acquire industrial site images of the target area and extract several key elements and element feature information of each key element from the industrial site images.

[0060] The generation module is used to obtain the communication node feature information of each communication node configured with a wireless communication module in the target area, construct a communication node knowledge graph of the target area based on the communication node feature information, and combine the element feature information with the communication node knowledge graph to generate an enhanced knowledge graph.

[0061] The prediction module is used to learn the graph structure data of the enhanced knowledge graph using a graph neural network with an attention mechanism, capture the deep-level associations between communication nodes and key elements, and obtain the communication quality prediction requirements of each communication node output by the graph neural network; wherein, the communication quality prediction requirements include several dimensions of communication requirement parameters.

[0062] The solution module is used to solve the communication resource configuration scheme for each communication node based on the communication quality requirement parameters of several dimensions in the communication quality prediction requirements and the influence mapping relationship between the communication quality requirement parameters and different communication resources.

[0063] The configuration module is used to perform communication resource configuration of the wireless communication module of each communication node within the area according to the communication resource configuration scheme, and realize industrial field communication based on the communication resource configuration results.

[0064] The beneficial effects of this invention are as follows: It proposes a high-quality communication method and system for wireless communication modules within a regional area. By extracting key elements and element feature information from industrial site images and integrating them into the communication node knowledge graph of the target area, an enhanced knowledge graph is formed. A graph neural network with an attention mechanism is used to learn the graph structure data of the enhanced knowledge graph, capturing the deep-level correlation between communication nodes and key elements to obtain predicted communication quality requirements. Then, according to the influence mapping relationship between communication quality requirement parameters and different communication resources, the communication resource configuration scheme for each communication node is solved, thereby executing industrial site communication. Thus, this invention accurately captures the dynamic changes and characteristics of equipment, personnel, and materials in industrial scenarios, and uses a combination of knowledge graph fusion and graph neural networks to achieve early prediction of communication demand changes. This solves the lag problem of traditional scheduling, achieves optimal resource allocation, ensures system communication stability and efficiency, and ultimately improves the overall operating efficiency of industrial automation systems. Attached Figure Description

[0065] Figure 1 This is a flowchart illustrating the high-quality communication method for wireless communication modules within the region of this invention.

[0066] Figure 2 This is a schematic diagram of the structure of a high-quality communication system for wireless communication modules within the region of this invention. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0068] This invention provides a high-quality communication method for wireless communication modules within a specific area, referring to... Figure 1 , Figure 1 This is a flowchart illustrating a high-quality communication method for a wireless communication module within a specified area according to an embodiment of the present invention.

[0069] In this embodiment, a high-quality communication method for a wireless communication module within a certain area is provided, the method comprising the following steps:

[0070] S1: Collect industrial site images of the target area, and extract several key elements and element feature information of each key element from the industrial site images;

[0071] S2: Obtain the communication node feature information of each communication node configured with a wireless communication module within the target area; construct a communication node knowledge graph of the target area based on the communication node feature information; combine the element feature information with the communication node knowledge graph to generate an enhanced knowledge graph.

[0072] S3: The graph structure data of the enhanced knowledge graph is learned by using a graph neural network with an attention mechanism to capture the deep-level association between communication nodes and key elements, and to obtain the communication quality prediction requirements of each communication node output by the graph neural network; wherein, the communication quality prediction requirements include several dimensions of communication requirement parameters.

[0073] S4: Based on the communication quality demand parameters of several dimensions in the communication quality prediction demand, and according to the influence mapping relationship between the communication quality demand parameters and different communication resources, solve the communication resource configuration scheme for each communication node;

[0074] S5: According to the communication resource configuration scheme, perform communication resource configuration for the wireless communication module of each communication node within the area, and realize industrial field communication based on the communication resource configuration results.

[0075] It should be noted that industrial sites are characterized by a variety of equipment, high personnel mobility, and dynamic material allocation. The communication needs of communication nodes (such as various sensors, control terminals, and gateways) change dynamically with the surrounding environment (personnel distribution, equipment operating status, material location, etc.). Traditional communication resource scheduling methods often rely on preset rules or passive feedback mechanisms based on network status, failing to fully explore the deep-seated relationships between communication nodes and key elements (equipment, personnel, materials, etc.) in the industrial scenario. This results in insufficient timeliness and accuracy of resource scheduling, making it difficult to adapt to the dynamically changing industrial environment, and consequently affecting communication quality and the overall operational efficiency of industrial automation systems.

[0076] To address the aforementioned issues, this embodiment extracts key elements and element feature information from industrial site images, integrates them into the communication node knowledge graph of the target area, forming an enhanced knowledge graph. A graph neural network incorporating an attention mechanism is then used to learn the graph structure data of this enhanced knowledge graph, capturing the deep-level relationships between communication nodes and key elements to obtain predicted communication quality requirements. Furthermore, based on the influence mapping relationship between communication quality requirement parameters and different communication resources, a communication resource configuration scheme for each communication node is solved, thereby executing industrial site communication. Thus, this invention, by accurately capturing the dynamic changes and characteristics of equipment, personnel, and materials in industrial scenarios and employing a combination of knowledge graph fusion and graph neural networks, achieves advance prediction of changes in communication requirements, ensuring the stability and efficiency of system communication.

[0077] In a preferred embodiment, the steps of acquiring industrial site images of the target area and extracting several key elements and element feature information of each key element from the industrial site images specifically include:

[0078] S11: Receive industrial site images collected and transmitted by several image acquisition devices deployed in the target area, and use image semantic segmentation to identify and extract several key elements and element features of each key element in the industrial site images.

[0079] S12: For each identified key element, extract the element feature vector from its element features to generate element feature information for each key element; wherein, the key element includes at least one of equipment element, personnel element, and material element, the element features include element location coordinates, element quantity, and element activity status, and the element feature vector includes feature parameters of several dimensions.

[0080] In this embodiment, industrial scene images collected and transmitted by distributed cameras within the target area are first received. Image semantic segmentation technology is used to accurately identify and extract key elements and element features in the images. Furthermore, element feature vectors are extracted to form structured element feature information. Through multi-device collaborative acquisition and semantic segmentation, the problem of incomplete identification caused by the large variety and scattered distribution of elements in industrial scenes is solved. At the same time, unstructured images are transformed into structured feature vectors to ensure that scene information can be efficiently integrated with communication node data, thereby improving the richness and accuracy of subsequent maps.

[0081] In practical applications, image semantic segmentation is used to identify and extract several key elements and the element features of each key element from industrial site images. The specific steps include:

[0082] S111: Input the received industrial site images into the U-Net architecture deep learning model, perform pixel-level classification on the images, and identify at least one key element among equipment elements, personnel elements, and material elements, as well as the element features of each key element;

[0083] The U-Net architecture deep learning model uses a loss function that combines cross-entropy loss, Dice loss, and smoothing loss during training, and its specific expression is as follows:

[0084] ;

[0085] In the formula, This represents the loss function used during training of the U-Net architecture deep learning model. Represents cross-entropy loss, This indicates Dice's loss. Indicates smoothing loss. , These are the weighting coefficients.

[0086] In this embodiment, the acquired industrial site images are input into the U-Net architecture deep learning model. The symmetrical encoder-decoder structure of U-Net achieves pixel-level classification, accurately identifying key elements and features such as equipment (e.g., robotic arms, AGVs), personnel (e.g., operators), and materials (e.g., vehicle parts, electronic component trays). Simultaneously, to improve model segmentation accuracy, a joint loss function is used during training. Cross-entropy loss optimizes the overall error of the classification task, Dice loss addresses the pixel imbalance between foreground (e.g., small materials) and background (e.g., the ground) in industrial scenes, and smoothing loss reduces jagged errors at segmentation boundaries, resulting in more accurate element edge recognition. Compared to traditional semantic segmentation models, U-Net has stronger pixel-level classification capabilities. Combined with the joint loss function, it solves the problem of missed identification of small elements due to class imbalance and optimizes segmentation boundary accuracy, significantly improving the accuracy of key element recognition and providing high-quality data for subsequent feature extraction.

[0087] In a preferred embodiment, the step of acquiring communication node feature information for each communication node configured with a wireless communication module within the target area, and constructing a communication node knowledge graph for the target area based on the communication node feature information, specifically includes:

[0088] S21: Collect the node attribute information of each communication node in the target area that is equipped with a wireless communication module in advance, and generate the node feature vector of each communication node as the communication node feature information.

[0089] S22: Construct edges between communication nodes based on physical connection associations, functional associations, and data flow associations, and calculate edge weights by combining communication frequency, communication data volume, and communication data dependency.

[0090] S23: Construct a knowledge graph of communication nodes for the target region based on the set of communication nodes consisting of all communication nodes, the set of edges consisting of all edges, and the set of edge weights consisting of all edge weights.

[0091] In practical applications, the expression for calculating edge weights is as follows:

[0092] ;

[0093] In the formula, This represents the edge weight between communication nodes i and j. This indicates the communication frequency between communication nodes i and j. This represents the amount of data communicated between communication nodes i and j. This represents the data dependency between communication nodes i and j. , , These are the weighting coefficients.

[0094] In this embodiment, node attribute information (such as device type, function, location coordinates, and communication protocol) of all communication nodes (such as sensors, control terminals, and gateways) configured with wireless communication modules within the target area is collected in advance. These attributes are then transformed into multi-dimensional node feature vectors as communication node feature information. Based on the actual associations between communication nodes, edges between nodes are constructed, including physical connection associations (such as the communication connection between sensors and gateways), functional associations (such as gateways relying on sensor data to implement control functions), and data flow associations (such as data flowing from sensors to gateways). The weight of each edge is calculated by combining communication frequency, communication data volume, and communication data dependency. Finally, the communication node set V, edge set E, and edge weight set W are integrated to construct a communication node knowledge graph that reflects the strength of node associations. .

[0095] It should be noted that when calculating edge weights, the communication frequency, communication data volume, and data dependency between the two communicating nodes are considered to set the appropriate weights. , , As a weighting coefficient (summing up to 1, it can be adjusted according to the needs of industrial scenarios, such as scenarios with high real-time requirements), (It can be set to 0.5 to enhance the influence of frequency), quantifying the information of the three dimensions of communication frequency (dynamic interaction strength), data volume (transmission load), and dependency (functional correlation tightness) into edge weights, accurately reflecting the correlation strength between nodes.

[0096] Therefore, by visualizing the dispersed communication nodes in the form of a graph, the strength of the association between nodes is clearly quantified, avoiding the blind resource scheduling caused by the ambiguity of node relationships in traditional methods, and providing a structured communication network model for subsequent integration with scene elements.

[0097] Based on this, the step of combining the element feature information with the communication node knowledge graph to generate an enhanced knowledge graph specifically includes:

[0098] S24: For the element feature vector corresponding to the element feature information Dimensionality reduction is performed, and the reduced element feature vectors are... Communication node feature vector corresponding to communication node feature information Perform feature concatenation to construct a fused feature vector. And perform normalization processing;

[0099] S25: Generate corresponding element nodes for each key element, and automatically identify and match communication nodes and element nodes that meet the matching rules according to the preset matching rules, and generate associated edges between communication nodes and element nodes that meet the matching rules.

[0100] S26: Based on the communication node feature vector of each communication node and the element feature vector of each key element Calculate the cosine similarity and Euclidean distance between each communication node and each key element, and calculate the weights according to the preset weight coefficients as the weights of the associated edges.

[0101] S27: Construct an enhanced knowledge graph of the target region based on the node set consisting of communication nodes and element nodes, the set of associated edges consisting of all associated edges, and the set of associated edge weights consisting of all associated edge weights.

[0102] In this embodiment, the element feature vectors are dimensionality reduced (e.g., using PCA algorithm) to avoid the curse of dimensionality caused by high-dimensional features. The dimensionality-reduced vectors are then concatenated with the communication node feature vectors to generate a fused feature vector. Normalization (e.g., Min-Max normalization) is used to unify the data scale. After this, corresponding element nodes are generated for each key element and added to the original communication node knowledge graph. Then, based on preset matching rules (e.g., the control range of the communication node covers the element node's location, and the element's function matches the communication node's function), communication nodes and element nodes are automatically identified and matched, generating association edges between the matched nodes. Finally, the association edge weights are calculated: the cosine similarity (measuring feature similarity, such as the compatibility between device type and control terminal) and Euclidean distance (measuring location proximity, such as the distance between a sensor and material) are calculated between the communication node and element node, and their weighted sum is calculated using preset weight coefficients as the association edge weights. Finally, the node sets of communication nodes and element nodes are integrated. The set of edges of primary edges and associated edges and the weight set of the original edge weight and the associated edge weight. Generate augmented knowledge graphs .

[0103] Therefore, this invention, through feature fusion and the construction of associated edges, enables the knowledge graph to not only contain communication network relationships, but also reflect the impact of scene dynamics on communication, providing rich structural data for subsequent graph neural networks to capture scenes and communication associations.

[0104] In a preferred embodiment, the step of learning the graph structure data of the enhanced knowledge graph using a graph neural network incorporating an attention mechanism to capture the deep-level relationships between communication nodes and key elements, and obtaining the communication quality prediction requirements for each communication node output by the graph neural network, specifically includes:

[0105] S31: Construct an initial multi-layer graph neural network, using historical communication quality parameters of the target area and key element feature data extracted from industrial site images of the same period as the training set, with the training objective of minimizing prediction error, and train the multi-layer graph neural network using the training set;

[0106] The specific expression for the training loss function is as follows:

[0107] ;

[0108] In the formula, To predict communication demand, For actual communication needs, i represents the i-th communication node, and N represents the total number of communication nodes. Represents the vector norm;

[0109] S32: Combine the fused feature vector at the current time step. As initial node features By utilizing a trained multi-layer graph neural network combined with an attention mechanism, and employing a layer-by-layer node feature propagation and aggregation approach, the features of the nodes in the previous layer are utilized. Deriving the node features of the next layer ;

[0110] The expression derived using a multi-layer graph neural network combined with an attention mechanism is as follows:

[0111] ;

[0112] ;

[0113] In the formula, For attention weights, For attention magnitude, ∥ represents the feature concatenation operation. For activation function, For the neighboring nodes of communication node i in the augmented knowledge graph connected by edges or associated edges, For the first Layer trainable weight matrix, For activation functions;

[0114] S33: Based on the final node features obtained in the last layer, predict the communication quality requirements of the output communication node in the next time step by performing linear transformation and activation on the feature vector of each communication node. .

[0115] In this embodiment, during the model training phase, a multi-layer graph neural network is constructed. The training set uses historical communication quality parameters of the target region (such as rate, latency, and packet loss rate over the past 6 months) and industrial site element features (such as equipment operating status and personnel distribution during the same period). The objective is to minimize the L2 norm between predicted and actual communication demands, and the model is trained until convergence. During the demand prediction phase, the fused feature vector at the current moment is used as the initial node features of the graph neural network. The trained model, combined with an attention mechanism, updates node features through layer-by-layer node feature propagation and aggregation. Layer node features Perform linear transformation and activation to output the multi-dimensional communication quality prediction requirements for each communication node at the next time step.

[0116] Therefore, by applying the attention mechanism, the model can focus on key associations (such as prioritizing the association between the control terminal and the corresponding device), avoiding the limitation of traditional graph neural networks that treat all neighbor nodes equally. By combining historical scenarios and communication data for training, the model can capture the potential patterns between scenario dynamics and communication needs, significantly improving the prediction accuracy of communication quality requirements.

[0117] In a preferred embodiment, the communication quality requirement parameters in the predicted communication quality requirements include: communication transmission rate, communication transmission delay, and communication packet loss rate; the communication resources include communication bandwidth parameters, communication power parameters, and communication frequency parameters. The steps for solving the communication resource configuration scheme for each communication node based on the communication quality requirement parameters and the influence mapping relationship between the communication quality requirement parameters and different communication resources specifically include:

[0118] S41: Based on the communication quality requirement parameters of several dimensions in the communication quality prediction requirements, and according to the influence mapping relationship between communication transmission rate, communication transmission delay and communication transmission packet loss rate and different communication resources, the calculation of communication bandwidth parameters, communication power parameters and communication frequency parameters is performed in sequence.

[0119] The specific expression for the mapping relationship between communication transmission rate and communication resources is as follows:

[0120] ;

[0121] In the formula, This represents the communication transmission rate of communication node i. This represents the spectral efficiency coefficient (determined by the wireless environment in industrial scenarios; for example, 1.25 is used in unobstructed scenarios). The communication bandwidth parameter represents the communication node i.

[0122] The specific expression for the mapping relationship between communication transmission delay and communication resources is as follows:

[0123] ;

[0124] In the formula, This represents the communication transmission rate of communication node i. This represents the communication transmission delay of communication node i. This represents the amount of data per frame for communication node i. This represents the power attenuation coefficient (affected by environmental interference; for example, 150 is used in multi-device scenarios). This represents the communication power parameter of communication node i. This indicates the physical layer base latency (determined based on the module type, e.g., 10ms).

[0125] The expression for the mapping relationship between packet loss rate and communication resources is as follows:

[0126] ;

[0127] In the formula, This represents the communication power parameter of communication node i. This represents the packet loss rate during communication transmission of communication node i. Indicates the base packet loss rate. , The fitting coefficients are determined by the type of equipment and are typically 0.3 or 0.2. Here are the communication transmission frequency parameters for communication node i. ) indicates frequency The level of cleanliness due to interference (values ​​range from 0 to 10, with higher values ​​for smaller interference).

[0128] S42: Generate a communication resource configuration scheme for each communication node based on the communication bandwidth parameters, communication power parameters, and communication frequency parameters of each communication node.

[0129] In this embodiment, based on the linear relationship between communication transmission rate and bandwidth, the communication bandwidth parameters under the communication quality prediction requirements are calculated. Then, based on the characteristic that communication transmission delay is jointly determined by data transmission time, power attenuation time, and physical layer base delay, the communication power under the communication quality prediction requirements is calculated based on the solved communication bandwidth parameters and the communication transmission delay under the communication quality prediction requirements. Finally, based on the characteristic that communication transmission packet loss rate is negatively correlated with power and frequency interference cleanliness, the communication frequency parameters under the communication quality prediction requirements are calculated based on the solved communication power and the communication transmission packet loss rate under the communication quality prediction requirements.

[0130] Therefore, this invention transforms abstract communication requirements into specific resource parameters through a quantified mapping relationship, avoiding resource waste (such as over-configuration of bandwidth) or insufficiency (such as excessive latency due to low power) caused by traditional experience-based configuration, and achieving precise resource matching.

[0131] In a preferred embodiment, according to the communication resource configuration scheme, communication resource configuration is performed for the wireless communication module of each communication node within the area. Based on the communication resource configuration results, industrial field communication steps are implemented, specifically including:

[0132] S51: Extract the communication bandwidth parameters, communication power parameters, and communication frequency parameters of each communication node recorded in the communication resource configuration scheme;

[0133] S52: Send the communication bandwidth parameters, communication power parameters, and communication frequency parameters of each communication node to the wireless communication module of the corresponding communication node, drive each communication node to configure the communication configuration parameters of the wireless communication module, and realize industrial field communication based on the configuration results.

[0134] In this embodiment, the core configuration parameters of each communication node—communication bandwidth, communication power, and communication frequency—are extracted from the communication resource configuration scheme to form a structured configuration instruction (e.g., JSON format: {"node_id":"C1","bandwidth":"8MHz","power":"4dBm","frequency":"2.4GHz"}). This instruction is then sent to the wireless communication module of the corresponding communication node via an industrial bus (e.g., RS485) or a wireless gateway. After receiving the instruction, the module automatically adjusts its internal parameters (e.g., adjusting the bandwidth, power, and frequency of the RF module). Once the configuration is complete, high-quality communication in the industrial field (e.g., transmission of equipment control commands and production data feedback) can be achieved, ultimately resulting in a stable improvement in the communication quality of the industrial field.

[0135] This invention proposes a high-quality communication method for wireless communication modules within a regional area. By fusing image semantic segmentation with knowledge graphs, it accurately captures the dynamic changes of equipment, personnel, and materials in industrial scenarios. Combined with a graph neural network using an attention mechanism, it enables advance prediction of communication needs, avoiding resource scheduling lags caused by traditional passive feedback. Simultaneously, it employs resource allocation based on quantized mapping relationships to achieve optimal matching of bandwidth, power, and frequency, reducing resource consumption when communication modules are idle or have low communication demand. In the long term, this can reduce energy costs and module wear in industrial systems.

[0136] On the other hand, dynamically adjusting the resource configuration of the wireless communication module enables on-demand configuration and precise adaptation, while minimizing interference with other nodes. This ensures that multi-node communication within the area is mutually non-interfering and transmits efficiently, guaranteeing that the overall communication quality indicators (rate, latency, packet loss rate) within the area meet the high requirements of industrial automation scenarios.

[0137] Reference Figure 2 , Figure 2 This is a schematic diagram of the structure of a high-quality communication system for a wireless communication module within a specified area according to an embodiment of the present invention.

[0138] like Figure 2 As shown, the high-quality communication system for wireless communication modules within a specific area proposed in this embodiment of the invention includes:

[0139] The acquisition module 10 is used to acquire industrial site images of the target area and extract several key elements and element feature information of each key element from the industrial site images.

[0140] The generation module 20 is used to obtain the communication node feature information of each communication node configured with a wireless communication module in the target area, construct a communication node knowledge graph of the target area based on the communication node feature information, and combine the element feature information with the communication node knowledge graph to generate an enhanced knowledge graph.

[0141] The prediction module 30 is used to learn the graph structure data of the enhanced knowledge graph using a graph neural network with an attention mechanism, capture the deep-level association between communication nodes and key elements, and obtain the communication quality prediction requirements of each communication node output by the graph neural network; wherein, the communication quality prediction requirements include several dimensions of communication requirement parameters.

[0142] The solution module 40 is used to solve the communication resource configuration scheme for each communication node based on the communication quality demand parameters of several dimensions in the communication quality prediction demand and the influence mapping relationship between the communication quality demand parameters and different communication resources.

[0143] The configuration module 50 is used to perform communication resource configuration of the wireless communication module of each communication node within the area according to the communication resource configuration scheme, and realize industrial field communication based on the communication resource configuration results.

[0144] Other embodiments or specific implementations of the high-quality communication system of the wireless communication module within the scope of this invention can be referred to the above-described method embodiments, and will not be repeated here.

[0145] It is understood that in the description of this specification, references to terms such as "one embodiment," "another embodiment," "other embodiments," or "first embodiment to Nth embodiment," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0146] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0147] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A high-quality communication method for a wireless communication module within a regional area, characterized in that, The method includes the following steps: Collect industrial site images of the target area, and extract several key elements and element feature information of each key element from the industrial site images; Obtain communication node feature information for each communication node configured with a wireless communication module within the target area; construct a communication node knowledge graph for the target area based on the communication node feature information; combine the element feature information with the communication node knowledge graph to generate an enhanced knowledge graph. A graph neural network incorporating an attention mechanism is used to learn the graph structure data of the enhanced knowledge graph, capturing deep-level relationships between communication nodes and key elements, and obtaining the communication quality prediction requirement for each communication node output by the graph neural network; wherein, the communication quality prediction requirement includes several dimensions of communication requirement parameters; specifically including: An initial multi-layer graph neural network is constructed, using historical communication quality parameters of the target area and key element feature data extracted from industrial field images of the same period as the training set. The training objective is to minimize the prediction error, and the multi-layer graph neural network is trained using the training set. The specific expression for the training loss function is as follows: ； In the formula, To predict communication demand, For actual communication needs, i represents the i-th communication node, and N represents the total number of communication nodes. Represents the vector norm; The fused feature vector at the current time As initial node features By utilizing a trained multi-layer graph neural network combined with an attention mechanism, and employing a layer-by-layer node feature propagation and aggregation approach, the features of the nodes in the previous layer are utilized. Deriving the node features of the next layer ; The expression derived using a multi-layer graph neural network combined with an attention mechanism is as follows: ； ； In the formula, For attention weights, For the sake of attention, For feature splicing operations, For activation function, For the neighboring nodes of communication node i in the augmented knowledge graph connected by edges or associated edges, For the first Layer trainable weight matrix, For activation functions; Based on the final node features obtained in the last layer, the communication quality requirements of the output communication node at the next time step are predicted by performing linear transformation and activation on the feature vector of each communication node. ; Based on several dimensions of communication quality requirement parameters in the predicted communication quality demand, and according to the influence mapping relationship between these parameters and different communication resources, the communication resource configuration scheme for each communication node is solved. The several dimensions of communication quality requirement parameters in the predicted communication quality demand include: communication transmission rate, communication transmission delay, and communication packet loss rate. The communication resources include communication bandwidth parameters, communication power parameters, and communication frequency parameters. Specifically, they include: Based on the communication quality requirement parameters of several dimensions in the communication quality prediction requirements, and according to the influence mapping relationship between communication transmission rate, communication transmission delay and communication transmission packet loss rate and different communication resources, the communication bandwidth parameter, communication power parameter and communication frequency parameter are calculated sequentially. The specific expression for the mapping relationship between communication transmission rate and communication resources is as follows: ； In the formula, This represents the communication transmission rate of communication node i. This represents the spectral efficiency coefficient. Let represent the communication bandwidth parameter of communication node i; where the expression for the mapping relationship between communication transmission delay and communication resources is as follows: ； In the formula, This represents the communication transmission rate of communication node i. This represents the communication transmission delay of communication node i. This represents the amount of data per frame for communication node i. Indicates the power attenuation coefficient. This represents the communication power parameter of communication node i. Indicates the physical layer base delay; The expression for the mapping relationship between packet loss rate and communication resources is as follows: ； In the formula, This represents the communication power parameter of communication node i. This represents the packet loss rate during communication transmission of communication node i. Indicates the base packet loss rate. , These are the fitting coefficients. Here are the communication transmission frequency parameters for communication node i. ) indicates frequency Interference with cleanliness; Based on the communication bandwidth parameters, communication power parameters, and communication frequency parameters of each communication node, a communication resource configuration scheme for each communication node is generated. According to the communication resource configuration scheme, the communication resource configuration of the wireless communication module of each communication node within the area is performed, and the industrial field communication is realized based on the communication resource configuration results.

2. The high-quality communication method for a wireless communication module within a specified area as described in claim 1, characterized in that, The steps of acquiring industrial site images of the target area and extracting several key elements and element feature information of each key element from the industrial site images specifically include: It receives industrial field images collected and transmitted by several image acquisition devices deployed in the target area, and uses image semantic segmentation to identify and extract several key elements and element features of each key element in the industrial field images. For each identified key element, extract the element feature vector from its element features to generate element feature information for each key element; wherein, the key element includes at least one of equipment element, personnel element, and material element, the element features include element location coordinates, element quantity, and element activity status, and the element feature vector includes feature parameters of several dimensions.

3. The high-quality communication method for a wireless communication module within a specified area as described in claim 2, characterized in that, The method employs image semantic segmentation to identify and extract several key elements and the element features of each key element from industrial site images. Specifically, this includes: The received industrial site images are input into the U-Net architecture deep learning model to perform pixel-level classification of the images and identify at least one key element among equipment elements, personnel elements, and material elements, as well as the element features of each key element. The U-Net architecture deep learning model uses a loss function that combines cross-entropy loss, Dice loss, and smoothing loss during training, and its specific expression is as follows: ； In the formula, This represents the loss function used during training of the U-Net architecture deep learning model. Represents cross-entropy loss, This indicates Dice's loss. Indicates smoothing loss. , These are the weighting coefficients.

4. The high-quality communication method for a wireless communication module within a specified area as described in claim 2, characterized in that, The steps of acquiring communication node feature information for each communication node configured with a wireless communication module within a target area, and constructing a communication node knowledge graph for the target area based on the communication node feature information, specifically include: The node attribute information of each communication node configured with a wireless communication module in the target area is collected in advance, and the node feature vector of each communication node is generated as the communication node feature information. Based on the physical connection association, functional association and data flow association between communication nodes, the edges between communication nodes are constructed, and the edge weights are calculated by combining the communication frequency, communication data volume and communication data dependency. A knowledge graph of communication nodes for the target region is constructed based on the set of all communication nodes, the set of all edges, and the set of all edge weights.

5. The high-quality communication method for a wireless communication module within a specified area as described in claim 4, characterized in that, The specific expression for calculating edge weights is as follows: ； In the formula, This represents the edge weight between communication nodes i and j. This indicates the communication frequency between communication nodes i and j. This represents the amount of data communicated between communication nodes i and j. This represents the data dependency between communication nodes i and j. , , These are the weighting coefficients.

6. The high-quality communication method for a wireless communication module within a specified area as described in claim 4, characterized in that, The step of combining the element feature information with the communication node knowledge graph to generate an enhanced knowledge graph specifically includes: The element feature vector corresponding to the element feature information Dimensionality reduction is performed, and the reduced element feature vectors are... Communication node feature vector corresponding to communication node feature information Perform feature concatenation to construct a fused feature vector. And perform normalization processing; For each key element, a corresponding element node is generated. Based on the preset matching rules between communication nodes and key elements, communication nodes and element nodes that match the matching rules are automatically identified and matched. Related edges are generated between communication nodes and element nodes that match the matching rules. Based on the communication node feature vector of each communication node and the element feature vector of each key element Calculate the cosine similarity and Euclidean distance between each communication node and each key element, and calculate the weights according to the preset weight coefficients as the weights of the associated edges. An enhanced knowledge graph of the target region is constructed based on the node set consisting of communication nodes and element nodes, the set of associated edges consisting of all associated edges, and the set of associated edge weights consisting of all associated edge weights.

7. The high-quality communication method for a wireless communication module within a specified area as described in claim 1, characterized in that, According to the communication resource configuration scheme, the communication resource configuration of the wireless communication module of each communication node within the area is performed. Based on the communication resource configuration results, the industrial field communication steps are implemented, specifically including: Extract the communication bandwidth parameters, communication power parameters, and communication frequency parameters of each communication node recorded in the communication resource configuration scheme; The communication bandwidth, communication power, and communication frequency parameters of each communication node are sent to the wireless communication module of the corresponding communication node, driving each communication node to configure the communication configuration parameters of the wireless communication module, and realizing industrial field communication based on the configuration results.

8. A high-quality communication system for a wireless communication module within a regional area, characterized in that, include: The acquisition module is used to acquire industrial site images of the target area and extract several key elements and element feature information of each key element from the industrial site images. The generation module is used to obtain the communication node feature information of each communication node configured with a wireless communication module in the target area, construct a communication node knowledge graph of the target area based on the communication node feature information, and combine the element feature information with the communication node knowledge graph to generate an enhanced knowledge graph. The prediction module is used to learn the graph structure data of the enhanced knowledge graph using a graph neural network incorporating an attention mechanism, capture the deep-level relationships between communication nodes and key elements, and obtain the communication quality prediction requirement for each communication node output by the graph neural network; wherein, the communication quality prediction requirement includes several dimensions of communication requirement parameters; specifically including: An initial multi-layer graph neural network is constructed, using historical communication quality parameters of the target area and key element feature data extracted from industrial field images of the same period as the training set. The training objective is to minimize the prediction error, and the multi-layer graph neural network is trained using the training set. The specific expression for the training loss function is as follows: ； In the formula, To predict communication demand, For actual communication needs, i represents the i-th communication node, and N represents the total number of communication nodes. Represents the vector norm; The fused feature vector at the current time As initial node features By utilizing a trained multi-layer graph neural network combined with an attention mechanism, and employing a layer-by-layer node feature propagation and aggregation approach, the features of the nodes in the previous layer are utilized. Deriving the node features of the next layer ; The expression derived using a multi-layer graph neural network combined with an attention mechanism is as follows: ； ； In the formula, For attention weights, For the sake of attention, For feature splicing operations, For activation function, For the neighboring nodes of communication node i in the augmented knowledge graph connected by edges or associated edges, For the first Layer trainable weight matrix, For activation functions; Based on the final node features obtained in the last layer, the communication quality requirements of the output communication node at the next time step are predicted by performing linear transformation and activation on the feature vector of each communication node. ; The solution module is used to solve for the communication resource configuration scheme of each communication node based on several dimensions of communication quality requirement parameters in the communication quality prediction requirements, and according to the influence mapping relationship between the communication quality requirement parameters and different communication resources. The several dimensions of communication quality requirement parameters in the communication quality prediction requirements include: communication transmission rate, communication transmission delay, and communication transmission packet loss rate; the communication resources include communication bandwidth parameters, communication power parameters, and communication frequency parameters. Specifically, it includes: Based on the communication quality requirement parameters of several dimensions in the communication quality prediction requirements, and according to the influence mapping relationship between communication transmission rate, communication transmission delay and communication transmission packet loss rate and different communication resources, the communication bandwidth parameter, communication power parameter and communication frequency parameter are calculated sequentially. The specific expression for the mapping relationship between communication transmission rate and communication resources is as follows: ； In the formula, This represents the communication transmission rate of communication node i. This represents the spectral efficiency coefficient. Let represent the communication bandwidth parameter of communication node i; where the expression for the mapping relationship between communication transmission delay and communication resources is as follows: ； In the formula, This represents the communication transmission rate of communication node i. This represents the communication transmission delay of communication node i. This represents the amount of data per frame for communication node i. Indicates the power attenuation coefficient. This represents the communication power parameter of communication node i. Indicates the physical layer base delay; The expression for the mapping relationship between packet loss rate and communication resources is as follows: ； In the formula, This represents the communication power parameter of communication node i. This represents the packet loss rate during communication transmission of communication node i. Indicates the base packet loss rate. , These are the fitting coefficients. Here are the communication transmission frequency parameters for communication node i. ) indicates frequency Interference with cleanliness; Based on the communication bandwidth parameters, communication power parameters, and communication frequency parameters of each communication node, a communication resource configuration scheme for each communication node is generated. The configuration module is used to perform communication resource configuration of the wireless communication module of each communication node within the area according to the communication resource configuration scheme, and realize industrial field communication based on the communication resource configuration results.

Images