Self-organizing linkage control method for intelligent distributed nodes of whole house

Abstract

The invention relates to the technical field of Internet of Things, and particularly discloses a whole-house intelligent distributed node self-organizing linkage control method, which comprises distributed node arrangement, self-organizing network design, communication architecture construction, transmission protocol design, node linkage control and fault detection. According to the scheme, distributed nodes are arranged, a self-organizing network is designed, the reliability and flexibility of the system are improved, and the data transmission efficiency is optimized by adopting an edge-cloud architecture; a transmission protocol is designed to realize efficient federated learning among nodes, delay is reduced through local model training, the response speed is improved, complex data is processed by using an edge layer, local bandwidth and computing resources are saved, and the operation efficiency of the whole system is improved; the self-organizing mapping is adopted to effectively reduce the dimension of the high-dimensional data, identify and cluster the nodes in the similar behavior mode, the stability and safety of the system are enhanced, the linkage condition is dynamically adjusted by considering the user habits and scenes at the same time, the personalized service is realized, and the user experience is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and in particular to a self-organizing linkage control method for whole-house intelligent distributed nodes. Background Art

[0002] The whole house intelligence realizes the interconnection, intercommunication and mutual control of smart homes through the Internet of Things technology. By arranging intelligent nodes in various areas of the room, linkage control between devices can be realized. The general node linkage is highly dependent on the stability of the network. When some nodes fail or the network fluctuates, instability will occur, affecting the reliability of linkage control; general edge nodes usually have limited computing power, and complex data processing will cause delays, affecting the real-time response and linkage effect of the system; general linkage control methods are too dependent on user initial settings, lack dynamic learning capabilities, and cannot adapt to changing user needs and behaviors. In addition, the feature selection is not comprehensive enough, resulting in an inability to accurately reflect the true status and behavior of the nodes. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a self-organizing linkage control method for whole-house intelligent distributed nodes. In view of the fact that general node linkage is highly dependent on network stability, instability will occur when some nodes fail or the network fluctuates, affecting the reliability of linkage control, this solution arranges distributed nodes and designs a self-organizing network to improve the reliability and flexibility of the system, realize real-time monitoring and processing of environmental data, and adopts an edge-cloud architecture to optimize data transmission efficiency. At the same time, it ensures the stability of the overall function when some nodes fail, and has good node expansion capabilities; in view of the fact that general edge nodes usually have limited computing power, complex data processing will cause delays, affecting the real-time response and linkage effect of the system, this solution designs a transmission protocol to implement node Efficient federated learning between nodes reduces latency and improves response speed through local model training, while protecting user privacy. It uses the edge layer to process complex data, saves local bandwidth and computing resources, and improves the operating efficiency of the overall system, so that the system can adapt to environmental changes in a timely manner. In view of the problem that the general linkage control method is too dependent on the user's initial settings, lacks dynamic learning capabilities, cannot adapt to the ever-changing user needs and behaviors, and the feature selection is not comprehensive enough, resulting in the inability to accurately reflect the true status and behavior of the nodes, this solution uses self-organizing mapping to effectively reduce the dimensionality of high-dimensional data, identify and cluster nodes with similar behavior patterns, automatically identify related nodes and abnormal nodes, enhance the stability and security of the system, and at the same time consider user habits and scenarios to dynamically adjust the linkage conditions to achieve personalized services and improve user experience.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides a self-organizing linkage control method for whole-house intelligent distributed nodes, the method comprising the following steps:

[0005] Step S1: Distributed node deployment: edge nodes are deployed in various areas of the house. Edge nodes are divided into mobile nodes and fixed nodes. Each edge node is equipped with sensors and has the ability to monitor environmental data, perform operations, and communicate.

[0006] Step S2: Self-organizing network design. Each edge node adaptively constructs the network topology according to the signal strength and distance of the neighboring nodes, automatically selects and joins the self-organizing network, and uses wireless communication for data transmission. The self-organizing network adopts a redundant design. When an individual node fails, it continues to communicate with other nodes through network reorganization.

[0007] Step S3: Communication architecture construction, constructing edge-cloud architecture, which consists of edge layer, network layer, cloud layer and application layer from bottom to top. Data is collected from the bottom layer and transmitted upward layer by layer. Users access services at the application layer.

[0008] Step S4: Design of transmission protocol: Design of transmission protocol to realize distributed computing of each node and federated learning between nodes, and regular update and optimization of local models to reflect the overall trend of the environment;

[0009] Step S5: Node linkage control, setting associated nodes, using self-organizing mapping to learn the association pattern between nodes, and automatically sending a linkage signal when a node is triggered to activate the associated node;

[0010] Step S6: Fault detection, using the node authentication mechanism to implement privacy protection for each edge node, real-time monitoring of the node operation status, and reporting to the system and neighboring nodes when a fault is detected. The system automatically adjusts the strategy to maintain the normal operation of other devices until the faulty node is repaired.

[0011] Furthermore, in step S3, the communication architecture is constructed, specifically including the following steps:

[0012] Step S31: The edge layer includes edge nodes and edge controllers. The edge nodes generate and store environmental data, create, process and train local models, and transmit the trained local models to the edge controllers. The edge controllers receive local models from multiple edge nodes and generate intermediate models after aggregation.

[0013] Step S32: The network layer establishes a communication channel, is responsible for realizing the communication between the edge layer and the cloud layer, sets the communication protocol for logical addressing, routing selection and congestion control, and maintains efficient data transmission between different layers;

[0014] Step S33: The cloud layer is provided with a cloud server, which receives the intermediate model of the edge layer and performs the computing task. The cloud server aggregates the uploaded intermediate model and generates a global model, and sends the global model back to each edge node through the network layer;

[0015] Step S34: The application layer provides application services to users, links smart homes and users' smart devices, and users directly access and manage each edge node and self-organizing network through application services and operation interfaces.

[0016] Furthermore, in step S4, the transmission protocol design specifically includes the following steps:

[0017] Step S41: Data generation, the user interacts with the environment, and the environment data is generated and stored at the edge node;

[0018] Step S42: Initial model creation, creating a randomly initialized neural network model at the edge node;

[0019] Step S43: Local training: the edge node trains the initial model, gradually updates the model parameters according to the environmental data, and generates a local model that reflects the characteristics of the node in the environment;

[0020] Step S44: Upload the local model. The edge node uploads the trained local model to the edge controller;

[0021] Step S45: edge aggregation, the edge controller aggregates the local models of each node to generate an intermediate model;

[0022] Step S46: The edge controller transmits the intermediate model to the cloud server;

[0023] Step S47: Cloud aggregation: The cloud server receives the intermediate models from each edge controller and generates a global model based on the data features learned in the training of each node;

[0024] Step S48: Transmitting the global model, the cloud server transmits the global model back to each edge controller;

[0025] Step S49: The edge controller transmits the global model back to each edge node to update the local model and perform further personalized training.

[0026] Furthermore, in step S5, the node linkage control specifically includes the following steps:

[0027] Step S51: data collection, collecting sensor data and environmental data from each edge node in each self-organizing network, and constructing a feature vector to represent the state and behavior of the node in each time period;

[0028] Step S52: association pattern learning, using a self-organizing map to create a SOM grid, mapping high-dimensional node features to a two-dimensional grid, and clustering the positions of similar nodes in the grid, including the following steps:

[0029] Step S521: Initialize a two-dimensional SOM grid and neurons, and the weight of each neuron is the same as the dimension of the node feature vector;

[0030] Step S522: Each time a feature vector of a node is input into the SOM grid, a neuron most similar to the feature vector is searched as the best matching unit of the node;

[0031] Step S523: iteratively update the weights of the best matching unit and adjacent neurons to make them close to the input feature vector until all nodes are mapped to the SOM grid;

[0032] Step S524: the SOM grid gradually converges, and a topological structure is formed after training is completed, and nodes with similar behavior patterns are mapped to similar grid positions;

[0033] Step S53: determining the associated nodes, searching for the associated nodes according to the node position distribution on the grid;

[0034] Step S54: association category division, dividing the nodes in each self-organizing network into different categories to represent different association modes, dividing each group of associated nodes into the same category, and activating the edge nodes in each category simultaneously when in use;

[0035] Step S55: abnormal node identification, finding abnormal nodes in the SOM grid that are far away from the main cluster, detecting abnormal device behaviors and association patterns, and issuing early warnings for abnormal nodes;

[0036] Step S56: Personalized adjustment: personalize the association mode based on the user's usage behavior and habits; perceive the environment in which the mobile node is located, dynamically adjust the linkage conditions, and automatically generate personalized linkage rules for different scenarios.

[0037] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0038] (1) In view of the high degree of dependence of general node linkage on network stability, instability will occur when some nodes fail or the network fluctuates, affecting the reliability of linkage control. This solution arranges distributed nodes and designs a self-organizing network to improve the reliability and flexibility of the system, realize real-time monitoring and processing of environmental data, and adopts an edge-cloud architecture to optimize data transmission efficiency. At the same time, it ensures the stability of the overall function when some nodes fail, and has good node expansion capabilities.

[0039] (2) In view of the fact that general edge nodes usually have limited computing power, complex data processing will cause delays, affecting the real-time response and linkage effect of the system, this solution designs a transmission protocol to achieve efficient federated learning between nodes, reduces delays and improves response speed through local model training, while protecting user privacy, using the edge layer to process complex data, saving local bandwidth and computing resources, and improving the operating efficiency of the overall system, so that the system can adapt to environmental changes in a timely manner.

[0040] (3) In view of the problem that the general linkage control method is too dependent on the user's initial settings, lacks dynamic learning capabilities, cannot adapt to the ever-changing user needs and behaviors, and is not comprehensive in feature selection, resulting in the inability to accurately reflect the true status and behavior of the nodes, this solution uses self-organizing mapping to effectively reduce the dimension of high-dimensional data, identify and cluster nodes with similar behavior patterns, automatically identify related nodes and abnormal nodes, enhance the stability and security of the system, and at the same time consider user habits and scenarios to dynamically adjust the linkage conditions, realize personalized services, and improve user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic diagram of a flow chart of a self-organizing linkage control method of whole-house intelligent distributed nodes proposed by the present invention;

[0042] Figure 2 The communication architecture diagram of step S3;

[0043] Figure 3 is a UML sequence diagram of the transmission protocol in step S4.

[0044] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0045] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0046] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0047] Example 1, see Figure 1 The present invention provides a self-organizing linkage control method for whole-house intelligent distributed nodes, the method comprising the following steps:

[0048] Step S1: Distributed node layout: edge nodes are arranged in various areas of the house. Edge nodes are divided into mobile nodes and fixed nodes. Fixed nodes include lights, curtains, air conditioners, and security equipment. Mobile nodes are users' smart terminal devices. Each edge node is equipped with sensors and has the ability to monitor environmental data, perform operations, and communicate.

[0049] Step S2: Self-organizing network design. Each edge node adaptively constructs the network topology according to the signal strength and distance of the neighboring nodes, automatically selects and joins the self-organizing network, and uses wireless communication for data transmission. The self-organizing network adopts a redundant design. When an individual node fails, it continues to communicate with other nodes through network reorganization.

[0050] Step S3: Communication architecture construction, constructing edge-cloud architecture, which consists of edge layer, network layer, cloud layer and application layer from bottom to top. Data is collected from the bottom layer and transmitted upward layer by layer. Users access services at the application layer.

[0051] Step S4: Design of transmission protocol: Design of transmission protocol to realize distributed computing of each node and federated learning between nodes, and regular update and optimization of local models to reflect the overall trend of the environment;

[0052] Step S5: Node linkage control, setting associated nodes, using self-organizing mapping to learn the association pattern between nodes, and automatically sending a linkage signal when a node is triggered to activate the associated node;

[0053] Step S6: Fault detection, using the node authentication mechanism to implement privacy protection for each edge node, real-time monitoring of the node operation status, and reporting to the system and neighboring nodes when a fault is detected. The system automatically adjusts the strategy to maintain the normal operation of other devices until the faulty node is repaired.

[0054] Example 2, see Figure 1This embodiment is based on the above embodiment. In step S2, when the edge node is a fixed node, an optimized communication path and protocol are preset to reduce the frequent routing update requirements, and the communication path is optimized for a long time to maintain information transmission efficiency; when the edge node is a mobile node, the network topology is adjusted in real time according to the location of the mobile node, the signal strength and the distance between the nodes. The mobile node continuously scans the signal, dynamically determines the signal strength and distance of the neighboring nodes, and reselects the best connection when necessary.

[0055] Example 3, see Figure 1 This embodiment is based on the above embodiment. In step S2, the self-organizing network adopts a redundant design as follows: path redundancy is adopted, each node maintains a connection with n adjacent nodes, and the network maintains n paths at the same time. When the main path fails, it immediately switches to the backup path; the AODV on-demand routing protocol is adopted, and the node initiates routing discovery when needed. When a node fails, an alternative path is quickly found through the AODV protocol.

[0056] Example 4, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S3, a communication architecture is constructed, which specifically includes the following steps:

[0057] Step S31: The edge layer includes edge nodes and edge controllers. The edge nodes generate and store environmental data, create, process and train local models, and transmit the trained local models to the edge controllers. The edge controllers receive local models from multiple edge nodes and generate intermediate models after aggregation.

[0058] Step S32: The network layer establishes a communication channel, which is responsible for realizing the communication between the edge layer and the cloud layer, setting the IP protocol for logical addressing, the BGP protocol for routing selection, and the TCP protocol for congestion control to maintain efficient data transmission between different layers;

[0059] Step S33: The cloud layer is provided with a cloud server, which receives the intermediate model of the edge layer and performs the computing task. The cloud server aggregates the uploaded intermediate model and generates a global model, and sends the global model back to each edge node through the network layer;

[0060] Step S34: The application layer provides application services to users, links smart homes and users' smart devices, and users directly access and manage each edge node and self-organizing network through application services and operation interfaces.

[0061] By executing the operations, the general node linkage is highly dependent on network stability, and instability will occur when some nodes fail or the network fluctuates, affecting the reliability of linkage control. This solution arranges distributed nodes and designs a self-organizing network to improve the reliability and flexibility of the system, realize real-time monitoring and processing of environmental data, and adopts an edge-cloud architecture to optimize data transmission efficiency. At the same time, it ensures the stability of the overall function in the event of some node failures, and has good node expansion capabilities.

[0062] Example 5, see Figure 1 , Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S4, the transmission protocol is designed, which specifically includes the following steps:

[0063] Step S41: Data generation: the user interacts with the environment, and the environment data is generated and stored at the edge node. The environment data includes device status, user behavior, and environment changes;

[0064] Step S42: Initial model creation, creating a randomly initialized neural network model at the edge node;

[0065] Step S43: Local training: the edge node trains the initial model, gradually updates the model parameters according to the environmental data, and generates a local model that reflects the characteristics of the node in the environment;

[0066] Step S44: Upload the local model. The edge node uploads the trained local model to the edge controller;

[0067] Step S45: edge aggregation, the edge controller aggregates the local models of each node to generate an intermediate model;

[0068] Step S46: The edge controller transmits the intermediate model to the cloud server;

[0069] Step S47: Cloud aggregation: The cloud server receives the intermediate models from each edge controller and generates a global model based on the data features learned in the training of each node;

[0070] Step S48: Transmitting the global model, the cloud server transmits the global model back to each edge controller;

[0071] Step S49: The edge controller transmits the global model back to each edge node to update the local model and perform further personalized training.

[0072] By executing the above operations, this solution designs a transmission protocol to achieve efficient federated learning between nodes, which can reduce delays and improve response speed through local model training, while protecting user privacy. It uses the edge layer to process complex data, saves local bandwidth and computing resources, and improves the operating efficiency of the overall system, so that the system can adapt to environmental changes in a timely manner.

[0073] Example 6, see Figure 1 This embodiment is based on the above embodiment. In step S5, the node linkage control specifically includes the following steps:

[0074] Step S51: data collection, collecting sensor data and environmental data from each edge node in each self-organizing network, and constructing a feature vector to represent the state and behavior of the node in each time period;

[0075] Step S52: association pattern learning, using a self-organizing map to create a SOM grid, mapping high-dimensional node features to a two-dimensional grid, and clustering the positions of similar nodes in the grid, including the following steps:

[0076] Step S521: Initialize a 10x10 two-dimensional SOM grid and neurons, and the weight of each neuron is the same as the dimension of the node feature vector;

[0077] Step S522: Each time a feature vector of a node is input into the SOM grid, a neuron most similar to the feature vector is searched as the best matching unit of the node;

[0078] Step S523: iteratively update the weights of the best matching unit and adjacent neurons to make them close to the input feature vector until all nodes are mapped to the SOM grid;

[0079] Step S524: the SOM grid gradually converges, and a topological structure is formed after training is completed, and nodes with similar behavior patterns are mapped to similar grid positions;

[0080] Step S53: determining the associated node, searching for the associated node according to the node position distribution on the grid, and if node A is adjacent to node B, taking node B as the associated node of node A;

[0081] Step S54: association category division, dividing the nodes in each self-organizing network into different categories for representing different association modes, dividing each group of associated nodes into the same category, and the nodes in each category are activated simultaneously when in use;

[0082] Step S55: abnormal node identification, finding abnormal nodes in the SOM grid that are far away from the main cluster, detecting abnormal device behaviors and association patterns, and issuing early warnings for abnormal nodes;

[0083] Step S56: Personalized adjustment: personalize the association mode based on the user's usage behavior and habits; perceive the environment in which the mobile node is located, dynamically adjust the linkage conditions, and automatically generate personalized linkage rules for different scenarios.

[0084] By executing the operations, the present invention aims to solve the problem that the general linkage control method is too dependent on the initial settings of the user, lacks dynamic learning ability, cannot adapt to the changing user needs and behaviors, and the feature selection is not comprehensive enough, resulting in the inability to accurately reflect the true status and behavior of the nodes. The present invention adopts self-organizing mapping to effectively reduce the dimension of high-dimensional data, identify and cluster nodes with similar behavior patterns, automatically identify related nodes and abnormal nodes, enhance the stability and security of the system, and dynamically adjust the linkage conditions considering user habits and scenarios, so as to realize personalized services and improve user experience.

[0085] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0086] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

[0087] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A self-organizing linkage control method for whole-house intelligent distributed nodes, characterized in that: The method comprises the following steps: Step S1: Distributed node deployment: edge nodes are deployed in various areas of the house. Edge nodes are divided into mobile nodes and fixed nodes. Each edge node is equipped with sensors and has the ability to monitor environmental data, perform operations, and communicate. Step S2: Self-organizing network design. Each edge node adaptively constructs the network topology according to the signal strength and distance of the neighboring nodes, automatically selects and joins the self-organizing network, and uses wireless communication for data transmission. The self-organizing network adopts a redundant design. When an individual node fails, it continues to communicate with other nodes through network reorganization. Step S3: Communication architecture construction, constructing edge-cloud architecture, which consists of edge layer, network layer, cloud layer and application layer from bottom to top. Data is collected from the bottom layer and transmitted upward layer by layer. Users access services at the application layer. Step S4: Design of transmission protocol: Design of transmission protocol to realize distributed computing of each node and federated learning between nodes, and regular update and optimization of local models to reflect the overall trend of the environment; Step S5: Node linkage control, setting associated nodes, using self-organizing mapping to learn the association pattern between nodes, and automatically sending a linkage signal when a node is triggered to activate the associated node; Step S6: Fault detection, using the node authentication mechanism to implement privacy protection for each edge node, real-time monitoring of the node operation status, and reporting to the system and neighboring nodes when a fault is detected. The system automatically adjusts the strategy to maintain the normal operation of other devices until the faulty node is repaired.

2. According to claim 1, a self-organizing linkage control method for whole-house intelligent distributed nodes is characterized in that: In step S3, the communication architecture is constructed, including the following steps: Step S31: The edge layer includes edge nodes and edge controllers. The edge nodes generate and store environmental data, create, process and train local models, and transmit the trained local models to the edge controllers. The edge controllers receive local models from multiple edge nodes and generate intermediate models after aggregation. Step S32: The network layer establishes a communication channel, is responsible for realizing the communication between the edge layer and the cloud layer, sets the communication protocol for logical addressing, routing selection and congestion control, and maintains efficient data transmission between different layers; Step S33: The cloud layer is provided with a cloud server, which receives the intermediate model of the edge layer and performs the computing task. The cloud server aggregates the uploaded intermediate model and generates a global model, and sends the global model back to each edge node through the network layer; Step S34: The application layer provides application services to users, links smart homes and users' smart devices, and users directly access and manage each edge node and self-organizing network through application services and operation interfaces.

3. According to claim 2, a self-organizing linkage control method for whole-house intelligent distributed nodes is characterized in that: In step S4, the transmission protocol design includes the following steps: Step S41: Data generation, the user interacts with the environment, and the environment data is generated and stored at the edge node; Step S42: Initial model creation, creating a randomly initialized neural network model at the edge node; Step S43: Local training: the edge node trains the initial model, gradually updates the model parameters according to the environmental data, and generates a local model that reflects the characteristics of the node in the environment; Step S44: Upload the local model. The edge node uploads the trained local model to the edge controller; Step S45: edge aggregation, the edge controller aggregates the local models of each node to generate an intermediate model; Step S46: The edge controller transmits the intermediate model to the cloud server; Step S47: Cloud aggregation: The cloud server receives the intermediate models from each edge controller and generates a global model based on the data features learned in the training of each node; Step S48: Transmitting the global model, the cloud server transmits the global model back to each edge controller; Step S49: The edge controller transmits the global model back to each edge node to update the local model and perform further personalized training.

4. The self-organizing linkage control method of the whole-house intelligent distributed nodes according to claim 1 is characterized in that: In step S5, the node linkage control includes the following steps: Step S51: data collection, collecting sensor data and environmental data from each edge node in each self-organizing network, and constructing a feature vector to represent the state and behavior of the node in each time period; Step S52: association pattern learning, using a self-organizing map to create a SOM grid, mapping high-dimensional node features to a two-dimensional grid, and clustering the positions of similar nodes in the grid, including the following steps: Step S521: Initialize a two-dimensional SOM grid and neurons, and the weight of each neuron is the same as the dimension of the node feature vector; Step S522: Each time a feature vector of a node is input into the SOM grid, a neuron most similar to the feature vector is searched as the best matching unit of the node; Step S523: iteratively update the weights of the best matching unit and adjacent neurons to make them close to the input feature vector until all nodes are mapped to the SOM grid; Step S524: the SOM grid gradually converges, and a topological structure is formed after training is completed, and nodes with similar behavior patterns are mapped to similar grid positions; Step S53: determining the associated nodes, searching for the associated nodes according to the node position distribution on the grid; Step S54: association category division, dividing the nodes in each self-organizing network into different categories to represent different association modes, dividing each group of associated nodes into the same category, and activating the edge nodes in each category simultaneously when in use; Step S55: abnormal node identification, finding abnormal nodes in the SOM grid that are far away from the main cluster, detecting abnormal device behaviors and association patterns, and issuing early warnings for abnormal nodes; Step S56: Personalized adjustment: personalize the association mode based on the user's usage behavior and habits; perceive the environment in which the mobile node is located, dynamically adjust the linkage conditions, and automatically generate personalized linkage rules for different scenarios.

Images