A power distribution internet of things multi-domain collaborative sensing method, device and medium

Abstract

The application discloses a power distribution Internet of Things multi-domain collaborative perception method and device and medium; relates to the field of power distribution Internet of Things, solves the problem that single-domain perception of a power distribution network cannot comprehensively understand the state of the entire power distribution network, establishes a multi-domain heterogeneous information network among a physical domain, an information domain, an energy domain and a social domain, effectively perceives the endogenous relationship between devices through different meta-paths, and finally aggregates semantic information in different dimensions through a graph attention network model, so that important feature attributes of the devices are obtained, the collaborative ability between interconnected power devices is improved, the problem that traditional perception methods excessively rely on single-domain knowledge and ignore massive multi-domain heterogeneous data in the power distribution Internet of Things, and cannot guarantee the differentiated device perception data service quality is solved.

Description

Technical Field

[0001] This application relates to the field of power distribution Internet of Things (IoT), and in particular to a multi-domain collaborative sensing method, device, and medium for power distribution IoT. Background Technology

[0002] With the continuous development of users and businesses, the number of access and control devices, as well as the amount of data, will continue to grow in the future. As the scale of the distribution network continues to expand, and the construction of smart distribution networks deepens and the integration of power distribution and consumption becomes more profound, the number of data acquisition terminals is increasing dramatically, and the frequency and scope of data acquisition are also greatly enhanced. The amount of power distribution and consumption data is developing from TB to PB levels, facing the challenges of effective integration, efficient storage, and high scalability of massive amounts of multi-source heterogeneous data. Power distribution business is gradually developing towards intelligence and lean management, which raises the need to further improve cross-business and cross-platform data analysis and processing capabilities.

[0003] Currently, the management of power distribution networks mainly relies on the design of single-domain knowledge perception algorithms to monitor and control the power system. However, different domains (physical domain, information domain, social domain, and energy domain) are independent, which limits the ability of single-domain perception to solve complex problems in smart grids. It can only monitor and perceive the equipment and conditions within each perception domain, and cannot provide a comprehensive understanding of the status of the entire power distribution network.

[0004] Therefore, it is evident that solving the problem that single-domain sensing of the distribution network cannot fully understand the status of the entire distribution network is a technical problem that urgently needs to be solved by those in this field. Summary of the Invention

[0005] The purpose of this application is to provide a multi-domain collaborative sensing method, device and medium for distribution Internet of Things (IoT) to solve the problem that single-domain sensing of distribution network cannot fully understand the status of the entire distribution network.

[0006] To address the aforementioned technical problems, this application provides a multi-domain collaborative sensing method for power distribution Internet of Things (IoT), comprising:

[0007] Obtain raw data from the physical domain, information domain, energy domain, and social domain of the power distribution Internet of Things (IoT).

[0008] Construct a multi-domain heterogeneous information network based on the original data;

[0009] In a multi-domain heterogeneous information network, a meta-path-based random walk is performed to obtain homogeneous sequences.

[0010] Discard irrelevant nodes in the homogeneous sequence except for power distribution equipment, and construct a homogeneous subgraph;

[0011] Train a graph attention network model based on homoproton graphs;

[0012] The trained graph attention network model is used to extract the feature attributes of power distribution equipment in the power distribution Internet of Things.

[0013] On the other hand, in the aforementioned multi-domain collaborative sensing method for the distribution IoT, a multi-domain heterogeneous information network is constructed based on the raw data, including:

[0014] Construct the first relationship graph between users based on the raw data from the social domain;

[0015] A second relationship graph between users and power distribution equipment is constructed based on the raw data from the physical domain and the raw data from the social domain.

[0016] Construct a third relationship graph between users and user information based on the raw data from the social domain and the raw data from the information domain;

[0017] Construct a fourth relationship diagram between power distribution equipment and power distribution equipment information based on the original data of the physical domain and the original data of the information domain;

[0018] By integrating the first, second, third, and fourth relationship diagrams, a multi-domain heterogeneous information network is obtained.

[0019] On the other hand, in the aforementioned multi-domain collaborative sensing method for the distribution IoT, a meta-path-based random walk is performed in a multi-domain heterogeneous information network to obtain homogeneous sequences, including:

[0020] The initial metapath is determined based on preset filtering criteria;

[0021] The transition probability of each node is obtained based on the type, order, and weights between nodes in the initial metapath;

[0022] Starting from the initial node of the initial metapath, a random walk is performed at each new node according to the transition probability until a preset termination condition is reached;

[0023] Store the homogeneous sequence generated by each meta-path walk.

[0024] On the other hand, in the aforementioned multi-domain collaborative sensing method for the distribution IoT, determining the initial meta-path based on preset screening conditions includes:

[0025] Obtain the average number of meta-path neighbors and the total number of nodes for each node;

[0026] Determine whether the average number of the number of the neighbors of each node on each metapath divided by the total number of nodes is greater than a preset threshold.

[0027] If so, then discard the current metapath;

[0028] If not, record the current metapath as the initial metapath.

[0029] On the other hand, in the aforementioned multi-domain collaborative sensing method for the distribution IoT, unrelated nodes other than power distribution equipment in the homogeneous sequence are discarded, and a homogeneous subgraph is constructed, including:

[0030] Discard irrelevant nodes in the homogeneous sequence except for power distribution equipment, and link nodes with the same power distribution equipment to obtain a homogeneous subgraph of power distribution equipment;

[0031] Based on the meta-path network in the homo-prime graph, calculate the co-occurrence frequency of each edge in the homo-prime graph;

[0032] Weights are assigned to edges in the homoprotic subgraph based on the co-occurrence frequency of each node.

[0033] On the other hand, in the aforementioned multi-domain collaborative sensing method for the distribution IoT, the feature attributes of power distribution equipment in the distribution IoT are extracted through a trained graph attention network model, including:

[0034] By inputting the homoproton graph into the trained graph attention network model for aggregation, the characteristic attributes of power distribution equipment in the power distribution Internet of Things are obtained.

[0035] On the other hand, in the aforementioned multi-domain collaborative sensing method for the Internet of Things for power distribution, the preset termination condition is a preset number of walking steps.

[0036] To address the aforementioned technical problems, this application also provides a multi-domain collaborative sensing device for power distribution Internet of Things (IoT), comprising:

[0037] The acquisition module is used to acquire raw data from the physical domain, information domain, energy domain, and social domain in the power distribution Internet of Things.

[0038] The building module is used to construct a multi-domain heterogeneous information network based on the raw data;

[0039] The sequence acquisition module is used to perform meta-path-based random walks in multi-domain heterogeneous information networks to acquire homogeneous sequences.

[0040] The module for constructing homogeneous subgraphs is used to discard irrelevant nodes in homogeneous sequences, excluding power distribution equipment, and construct homogeneous subgraphs.

[0041] The training model module is used to train a graph attention network model based on homoprime graphs.

[0042] The analysis module is used to extract the feature attributes of power distribution equipment in the power distribution Internet of Things through a trained graph attention network model.

[0043] To address the aforementioned technical problems, this application also provides a multi-domain collaborative sensing device for power distribution Internet of Things (IoT), comprising:

[0044] Memory, used to store computer programs;

[0045] The processor is used to implement the steps of the multi-domain collaborative sensing method of the power distribution Internet of Things when executing computer programs.

[0046] To address the aforementioned technical problems, this application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the aforementioned multi-domain collaborative sensing method for power distribution Internet of Things.

[0047] The multi-domain collaborative sensing method for the distribution IoT provided in this application acquires raw data from the physical, information, energy, and social domains of the distribution IoT; constructs a multi-domain heterogeneous information network based on the raw data; performs a meta-path-based random walk in the multi-domain heterogeneous information network to obtain homogeneous sequences among distribution devices; discards irrelevant nodes other than distribution devices in the homogeneous sequences to construct a homogeneous subgraph; trains a graph attention network model based on the homogeneous subgraph; and extracts the feature attributes of distribution devices in the distribution IoT through the trained graph attention network model. By establishing a multi-domain heterogeneous information network among the physical, information, energy, and social domains, and using different meta-paths, the intrinsic relationships between devices are effectively perceived. Finally, the graph attention network model aggregates semantic information from different dimensions to obtain important feature attributes of devices, thereby improving the collaborative capabilities between interconnected power devices. This solves the problem that traditional sensing methods rely too much on single-domain knowledge, ignore the massive multi-domain heterogeneous data in the distribution IoT, and cannot guarantee the quality of differentiated device sensing data services.

[0048] In addition, this application also provides an apparatus and a medium that correspond to the above method and have the same effect. Attached Figure Description

[0049] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 A flowchart illustrating a multi-domain collaborative sensing method for a power distribution Internet of Things (IoT) provided in this application embodiment;

[0051] Figure 2 This is a schematic diagram of a multi-domain heterogeneous information network mentioned in an embodiment of this application;

[0052] Figure 3 A data flow diagram of a multi-domain collaborative sensing method for power distribution Internet of Things provided in this application embodiment;

[0053] Figure 4A structural diagram of a multi-domain collaborative sensing device for power distribution Internet of Things provided in this application embodiment;

[0054] Figure 5 This is a structural diagram of another multi-domain collaborative sensing device for power distribution Internet of Things provided in an embodiment of this application. Detailed Implementation

[0055] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0056] The core of this application is to provide a multi-domain collaborative sensing method, device, and medium for power distribution Internet of Things.

[0057] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] Single-domain sensing in a distribution network refers to dividing the distribution network into several independent sensing domains when monitoring and sensing it within a smart grid. Each sensing domain is responsible for monitoring and managing the power distribution equipment and electricity consumption information within its corresponding area. It can achieve multiple functions such as real-time monitoring, fault location, and load management. However, single-domain sensing can only monitor and sense the equipment and conditions within each sensing domain, and cannot provide a comprehensive understanding of the entire distribution network's status. If a fault or anomaly involves multiple sensing domains, it may not be possible to immediately detect and accurately locate the problem.

[0059] Therefore, in order to address the problem that single-domain sensing of a distribution network cannot provide a comprehensive understanding of the entire distribution network's status, this embodiment provides a multi-domain collaborative sensing method for the Internet of Things (IoT) of distribution networks. Figure 1 A flowchart of a multi-domain collaborative sensing method for power distribution Internet of Things (IoT) provided in this application embodiment is shown below. Figure 1 As shown, it includes:

[0060] S11: Obtain the raw data from the physical domain, information domain, energy domain, and social domain in the power distribution Internet of Things;

[0061] S12: Construct a multi-domain heterogeneous information network based on the original data;

[0062] S13: Perform a meta-path-based random walk in a multi-domain heterogeneous information network to obtain homogeneous sequences;

[0063] S14: Discard irrelevant nodes in the homogeneous sequence except for power distribution equipment, and construct a homogeneous subgraph;

[0064] S15: Train a graph attention network model based on the homoproton graph;

[0065] S16: Extract the feature attributes of power distribution equipment in the power distribution Internet of Things through the trained graph attention network model.

[0066] In this embodiment, the physical domain refers to various power distribution devices in the power distribution Internet of Things, including smart power distribution terminals, transformers, etc.

[0067] Information domain: includes social domain user attribute space such as user type, age, social class, interests and hobbies, and physical domain device attribute space such as device type, location, environment, model, etc.

[0068] Energy domain: refers to the energy interaction information between social domain users and power distribution equipment;

[0069] Social domain: Users, as entities in the social domain, are access terminals for power distribution IoT devices. This includes users' social relationships and social information.

[0070] This embodiment does not limit the data representation format in various fields; it can be set according to actual needs. Preferably, the raw data is preprocessed, including steps such as data cleaning, feature extraction, and data transformation, to convert the data into a format that the network can process.

[0071] Figure 2 This is a schematic diagram of a multi-domain heterogeneous information network mentioned in an embodiment of this application; as shown... Figure 2 As shown, a multi-domain heterogeneous information network is constructed using raw data. The structure of this network includes defining the attributes of nodes and connections, as well as the connection methods between nodes. In this embodiment, nodes can represent users, devices, and locations in different domains, and edges represent connections between different domains.

[0072] Step S13 involves performing a meta-path-based random walk in a multi-domain heterogeneous information network to obtain homogeneous sequences. Meta-path-based random walks are a network analysis method used to uncover relationships between nodes. It simulates node behavior in a network by defining different types of paths or trajectories. During the random walk, each move is determined based on a specific probability, which can be based on the node's neighboring nodes or the type of the path. By simulating and statistically analyzing a large number of random walks, information such as the similarity between nodes and the importance of paths can be obtained. A homogeneous sequence is an ordered sequence of nodes with the same node type; in this embodiment, the homogeneous sequence is an ordered sequence of power distribution equipment nodes.

[0073] By exploring different meta-paths, the intrinsic connections between nodes are investigated. A random walk method based on meta-paths is used to capture different semantic relationships in multi-domain heterogeneous information networks, thereby obtaining homogeneous sequences with multi-domain endogenous information. The meta-paths mentioned in this embodiment refer to meta-paths between power distribution devices, such as: {Power Distribution Device-User-User-Power Distribution Device}, {Power Distribution Device-User-Power Distribution Device}, {Power Distribution Device-Location-Power Distribution Device}, {Power Distribution Device-Type-Power Distribution Device}, etc. A meta-path M is defined as a path defined on the network pattern Tg, denoted as... It represents an A1 to A1 l+1 The complex relationship between them is C = C1oC2o...oC n Where o represents the conformity operator between relations, A n Represents node type, C n Indicates the relation type.

[0074] Step S14 discards irrelevant nodes in the homogeneous sequence except for power distribution equipment, and constructs a homogeneous subgraph. By retaining only nodes related to the homogeneous sequence, the size of the subgraph can be reduced and the efficiency and accuracy of subsequent analysis can be improved. Based on the selected relevant nodes, a homogeneous subgraph is constructed. The homogeneous subgraph only contains nodes related to the homogeneous sequence and the edges between them.

[0075] Graph Attention Network (GAT) is a deep learning model for processing graph data. It introduces an attention mechanism to achieve information transfer and feature representation between nodes. The core idea of ​​the GAT model is to introduce an attention coefficient for each pair of connections between nodes, used to calculate the weights between nodes. This allows each node to focus on relevant neighboring nodes based on its correlation with other nodes.

[0076] Step S15: Training a graph attention network model based on the homoproton graph involves using the homoproton graph of the power distribution equipment as input to the graph attention network. It learns node representations from the homoproton graph data by utilizing node features, connections between nodes, and attention mechanisms, aggregating semantic information from different dimensions to improve the understanding and differentiation of different types of nodes in the network, thereby achieving the perception of the intrinsic connections between nodes in multiple domains. The loss function uses the mean squared error loss function, and the Adaptive Moment Estimation (Adam) optimization method is employed for training. The training speed and effectiveness are controlled by adjusting parameters such as the learning rate. During training, the number of neighbors of a node may vary greatly. When performing global computation on all nodes, nodes with a large number of neighbors will consume a lot of computational resources, while nodes with a small number of neighbors will be ignored. Neighbor sampling can balance the distribution of the number of neighbors, making the importance of each node more balanced, better preserving the characteristics of the graph structure, making the model more robust, and further improving the model's generalization ability.

[0077] S16: Extract the feature attributes of power distribution equipment in the power distribution Internet of Things through the trained graph attention network model.

[0078] The core idea of ​​GAT is to use an attention mechanism to dynamically calculate the importance weights between nodes and then aggregate node features based on these weights. Finally, a graph attention network is used to aggregate semantic information from different dimensions to obtain important device feature attributes, thereby improving the collaborative capabilities between interconnected power devices.

[0079] The multi-domain collaborative sensing method for the distribution IoT provided in this application acquires raw data from the physical, information, energy, and social domains of the distribution IoT; constructs a multi-domain heterogeneous information network based on the raw data; performs a meta-path-based random walk in the multi-domain heterogeneous information network to obtain homogeneous sequences among distribution devices; discards irrelevant nodes other than distribution devices in the homogeneous sequences to construct a homogeneous subgraph; trains a graph attention network model based on the homogeneous subgraph; and extracts the feature attributes of distribution devices in the distribution IoT through the trained graph attention network model. By establishing a multi-domain heterogeneous information network among the physical, information, energy, and social domains, and using different meta-paths, the intrinsic relationships between devices are effectively perceived. Finally, the graph attention network model aggregates semantic information from different dimensions to obtain important feature attributes of devices, thereby improving the collaborative capabilities between interconnected power devices. This solves the problem that traditional sensing methods rely too much on single-domain knowledge, ignore the massive multi-domain heterogeneous data in the distribution IoT, and cannot guarantee the quality of differentiated device sensing data services.

[0080] According to the above embodiments, in another embodiment, the multi-domain collaborative sensing method for the power distribution Internet of Things (IoT) constructs a multi-domain heterogeneous information network based on the original data, including:

[0081] Construct the first relationship graph between users based on the raw data from the social domain;

[0082] A second relationship graph between users and power distribution equipment is constructed based on the raw data from the physical domain and the raw data from the social domain.

[0083] Construct a third relationship graph between users and user information based on the raw data from the social domain and the raw data from the information domain;

[0084] Construct a fourth relationship diagram between power distribution equipment and power distribution equipment information based on the original data of the physical domain and the original data of the information domain;

[0085] By integrating the first, second, third, and fourth relationship diagrams, a multi-domain heterogeneous information network is obtained.

[0086] Based on the raw data of the social domain, a first relationship graph between users is constructed to build a user social relationship graph by utilizing the friend relationships of electricity users in the social domain.

[0087] A second relationship graph between users and power distribution equipment is constructed based on the raw data from the physical domain and the raw data from the social domain. This is to establish a relationship graph between social domain users and physical domain power distribution equipment based on the electricity consumption data of power distribution equipment and the electricity consumption behavior data of users collected from the energy domain.

[0088] Based on the raw data from the social domain and the raw data from the information domain, a third relationship graph between users and user information is constructed to build a social-information relationship graph by utilizing the user's friend relationships in the social domain and the user's location, age, and gender in the information domain.

[0089] Based on the original data in the physical domain and the original data in the information domain, a fourth relationship diagram is constructed between power distribution equipment and power distribution equipment information. This is to establish a relationship diagram between equipment and related information by using the equipment name in the physical domain and the equipment type, location, temperature, and other attributes in the information domain.

[0090] Integrating the first, second, third, and fourth relationship graphs yields a multi-domain heterogeneous information network. This integration of relationship graphs constructs a complex heterogeneous social power information network with multi-domain heterogeneous data collaboration, defined as... It contains a set of nodes V, a set of connections E, and a set of attribute values ​​W on the connections. The network also contains node mappings Φ:V→A, connection mapping node sequences Ψ:E→C, and attribute value type mappings θ:W→W, where A is a predefined node type, C is a predefined set of connection types, and W is a predefined set of attribute values.

[0091] According to the above embodiments, in another embodiment, the multi-domain collaborative sensing method of the above-mentioned power distribution Internet of Things performs a meta-path-based random walk in a multi-domain heterogeneous information network to obtain homogeneous sequences, including:

[0092] Determine the initial meta-path between power distribution equipment based on preset screening criteria;

[0093] The transition probability of each node is obtained based on the type, order, and weights between nodes in the initial metapath;

[0094] Starting from the initial node of the initial metapath, a random walk is performed at each new node according to the transition probability until a preset termination condition is reached;

[0095] Store the homogeneous sequence generated by each meta-path walk.

[0096] The initial meta-path is determined based on preset filtering criteria; in this embodiment, it is to filter meta-paths with electrical equipment as the starting and ending nodes. The initial meta-path is defined. It is a path defined on a multi-domain heterogeneous information network Tg, denoted as... It represents an A1 to A1 l+1 The complex relationship between them is C = C1oC2o...oC n Where o represents the conformity operator between relations, A n Represents node type, C n Indicates the relationship type. Based on the multi-domain data characteristics of the power distribution IoT, the meta-path is determined, including {Power Distribution Equipment-User-User-Power Distribution Equipment}, {Power Distribution Equipment-User-Power Distribution Equipment}, {Power Distribution Equipment-Location-Power Distribution Equipment}, {Power Distribution Equipment-Type-Power Distribution Equipment}, etc.

[0097] Construct a transition probability matrix. Based on the required meta-path, construct a transition probability matrix P between nodes. The transition probability p between nodes is affected by the type and order of nodes in the meta-path, as well as the weights between nodes. The formula for the transition probability p is as follows:

[0098]

[0099] The node of type t in step i is represented as v. t i The neighboring nodes of node type t+1 are represented as v. i+1 The attribute value of a neighbor node that meets the meta-path condition is represented as w. vtv i+1 w viu This represents the attribute values ​​of all neighboring nodes that meet the metapath criteria.

[0100] Starting from the initial node, a random walk is performed on the heterogeneous social power information network constructed in S12 according to the transition probabilities in the probability matrix. After reaching a new node, selection continues according to the transition probabilities, and this process is repeated until a preset termination condition is met; specifically, the preset termination condition is: a preset number of walk steps. The homogeneous sequence generated by each meta-path walk is stored.

[0101] According to the above embodiments, in another embodiment, the determination of the initial meta-path based on preset filtering conditions in the above-described multi-domain collaborative sensing method for power distribution IoT includes:

[0102] Obtain the average number of meta-path neighbors and the total number of nodes for each node;

[0103] Determine whether the average number of the number of the neighbors of each node on each metapath divided by the total number of nodes is greater than a preset threshold.

[0104] If so, then discard the current metapath;

[0105] If not, record the current metapath as the initial metapath.

[0106] For each initial metapath selection, the average number of metapath neighbors for each node is calculated. If an average divided by the total number of nodes is greater than a given threshold, the metapath selection is abandoned.

[0107] According to the above embodiments, in another embodiment, the multi-domain collaborative sensing method for the power distribution Internet of Things discards irrelevant nodes other than power distribution equipment in the homogeneous sequence and constructs a homogeneous subgraph, including:

[0108] Discard irrelevant nodes in the homogeneous sequence except for power distribution equipment, and link nodes with the same power distribution equipment to obtain a homogeneous subgraph of power distribution equipment;

[0109] Based on the meta-path network in the homo-prime graph, calculate the co-occurrence frequency of each edge in the homo-prime graph;

[0110] Weights are assigned to edges in the homoprotic subgraph based on the co-occurrence frequency of each node.

[0111] By discarding irrelevant nodes and using the node sequence obtained in S13, the relationships between nodes of the same power distribution equipment are linked, with the goal of constructing a homoproton graph. The homoproton graph structure is defined as follows. The set of nodes Adjacency matrix and the weights between nodes By statistically analyzing the frequency of node occurrence under different metapaths and assigning weights to each edge, the topological structure of the homoproton graph of the power distribution equipment is finally obtained.

[0112] According to the above embodiments, in another embodiment, the multi-domain collaborative sensing method for the power distribution Internet of Things (IoT) extracts the feature attributes of power distribution equipment in the IoT through a trained graph attention network model, including:

[0113] By inputting the homoproton graph into the trained graph attention network model for aggregation, the characteristic attributes of power distribution equipment in the power distribution Internet of Things are obtained.

[0114] By using the homoprotic graph of the power distribution equipment in S14 as input to the graph attention network model, the model learns node representations in the homoprotic graph data by utilizing node features, connections between nodes, and attention mechanisms. This aggregates semantic information from different dimensions, improving the understanding and differentiation of different types of nodes in the network, thereby achieving the perception of the intrinsic connections between nodes in multiple domains. The loss function utilizes the mean squared error loss function and is optimized using the Adaptive Moment Estimation (Adam) method. Training is performed by adjusting parameters such as the learning rate to control the training speed and effectiveness. During training, the number of neighbors of a node may vary greatly. When performing global computation on all nodes, nodes with more neighbors will consume a large amount of computational resources, while nodes with fewer neighbors will be ignored. Neighbor sampling can balance the distribution of the number of neighbors, making the importance of each node more balanced, better preserving the characteristics of the graph structure, making the model more robust, and further improving the model's generalization ability.

[0115] The homoproton graph is input into the trained graph attention network model for aggregation, perceiving the intrinsic relationships between multi-domain information of the device, and obtaining the important feature attribute output Z of the power distribution equipment. i It can be further used for user power load forecasting tasks in distribution networks.

[0116] In order to enable those skilled in the art to better understand this solution, Figure 3 This application provides a data flow diagram of a multi-domain collaborative sensing method for power distribution IoT. It describes how a homogeneous sequence is obtained from a multi-domain heterogeneous information network based on random walks of meta-paths, further constructing a homogeneous subgraph, and using a graph attention network model to perceive node relationships, thereby obtaining the characteristic attributes Z of the power distribution equipment. i .

[0117] The multi-domain collaborative sensing method for power distribution IoT has been described in detail in the above embodiments. This application also provides embodiments corresponding to the multi-domain collaborative sensing device for power distribution IoT. It should be noted that this application describes the embodiments of the device part from two perspectives: one is based on functional modules, and the other is based on hardware.

[0118] From the perspective of functional modules Figure 4A structural diagram of a multi-domain collaborative sensing device for power distribution Internet of Things (IoT) provided in this application embodiment is shown below. Figure 4 As shown, the multi-domain collaborative sensing device for the power distribution Internet of Things includes:

[0119] The acquisition module 21 is used to acquire raw data from the physical domain, information domain, energy domain, and social domain in the power distribution Internet of Things.

[0120] Module 22 is used to construct a multi-domain heterogeneous information network based on the raw data;

[0121] Sequence acquisition module 23 is used to perform meta-path-based random walks in multi-domain heterogeneous information networks to acquire homogeneous sequences;

[0122] The module 24 for constructing homogeneous subgraphs is used to discard irrelevant nodes in the homogeneous sequence except for power distribution equipment and construct homogeneous subgraphs.

[0123] Training model module 25 is used to train a graph attention network model based on homoprime graphs;

[0124] Analysis module 26 is used to extract the feature attributes of power distribution equipment in the power distribution Internet of Things through the trained graph attention network model.

[0125] This embodiment utilizes a multi-domain collaborative sensing device for the power distribution Internet of Things (IoT). The acquisition module 21 acquires raw data from the physical, information, energy, and social domains of the IoT. The construction module 22 constructs a multi-domain heterogeneous information network based on the raw data. The sequence acquisition module 23 performs a random walk based on meta-paths within the multi-domain heterogeneous information network to acquire homogeneous sequences. The homogeneous subgraph construction module 24 discards irrelevant nodes (excluding power distribution equipment) from the homogeneous sequences and constructs a homogeneous subgraph. The model training module 25 trains a graph attention network model based on the homogeneous subgraph. The analysis module 26 extracts the characteristic attributes of power distribution equipment in the IoT using the trained graph attention network model. By establishing a multi-domain heterogeneous information network across the physical, information, energy, and social domains, and using different meta-paths, the intrinsic relationships between devices are effectively perceived. Finally, the graph attention network model aggregates semantic information from different dimensions to obtain important characteristic attributes of the devices, improving the collaborative capabilities between interconnected power devices. This addresses the problem that traditional sensing methods rely too heavily on single-domain knowledge, neglecting the massive amounts of multi-domain heterogeneous data in the IoT, and failing to guarantee the quality of differentiated device sensing data services.

[0126] Since the embodiments of the apparatus and the embodiments of the method correspond to each other, please refer to the description of the embodiments of the method for the embodiments of the apparatus, which will not be repeated here.

[0127] Figure 5A structural diagram of another multi-domain collaborative sensing device for power distribution Internet of Things provided in this application embodiment is shown below. Figure 5 As shown, the multi-domain collaborative sensing device of the power distribution Internet of Things includes: a memory 30 for storing computer programs;

[0128] The processor 31 is used to execute a computer program to implement the steps of the method for obtaining user operation habit information as described in the above embodiment (multi-domain collaborative sensing method for power distribution Internet of Things).

[0129] The multi-domain collaborative sensing device for the power distribution Internet of Things provided in this embodiment may include, but is not limited to, smartphones, tablets, laptops, or desktop computers.

[0130] The processor 31 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 31 may be implemented using at least one of the following hardware forms: Digital Signal Processor (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 31 may also include a main processor and a coprocessor. The main processor, also known as the Central Processing Unit (CPU), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 31 may integrate a Graphics Processing Unit (GPU), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 31 may also include an Artificial Intelligence (AI) processor, which handles computational operations related to machine learning.

[0131] The memory 30 may include one or more computer-readable storage media, which may be non-transitory. The memory 30 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In this embodiment, the memory 30 is used to store at least the following computer program 301, which, after being loaded and executed by the processor 31, is capable of implementing the relevant steps of the multi-domain collaborative sensing method for the distribution IoT disclosed in any of the foregoing embodiments. In addition, the resources stored in the memory 30 may also include an operating system 302 and data 303, and the storage method may be temporary or permanent storage. The operating system 302 may include Windows, Unix, Linux, etc. The data 303 may include, but is not limited to, the data involved in implementing the multi-domain collaborative sensing method for the distribution IoT.

[0132] In some embodiments, the multi-domain collaborative sensing device of the power distribution Internet of Things may further include a display screen 32, an input / output interface 33, a communication interface 34, a power supply 35, and a communication bus 36.

[0133] Those skilled in the art will understand that Figure 5 The structure shown does not constitute a limitation on multi-domain collaborative sensing devices for the Internet of Things for distribution, and may include more or fewer components than shown.

[0134] The multi-domain collaborative sensing device for power distribution IoT provided in this application includes a memory and a processor. When the processor executes the program stored in the memory, it can implement the following method: a multi-domain collaborative sensing method for power distribution IoT, which acquires raw data of the physical domain, information domain, energy domain, and social domain in the power distribution IoT respectively; constructs a multi-domain heterogeneous information network based on the raw data; performs a meta-path-based random walk in the multi-domain heterogeneous information network to acquire homogeneous sequences among power distribution devices; discards irrelevant nodes other than power distribution devices in the homogeneous sequences to construct a homogeneous subgraph; trains a graph attention network model based on the homogeneous subgraph; and extracts the feature attributes of power distribution devices in the power distribution IoT through the trained graph attention network model. By establishing a multi-domain heterogeneous information network among the physical, information, energy, and social domains, and effectively perceiving the intrinsic relationships between devices through different meta-paths, and finally aggregating semantic information from different dimensions through a graph attention network model, we can obtain important feature attributes of devices, improve the collaborative capabilities between interconnected power devices, and solve the problem that traditional perception methods rely too much on single-domain knowledge, ignore the massive multi-domain heterogeneous data in the power distribution Internet of Things, and cannot guarantee the quality of differentiated device perception data services.

[0135] Finally, this application also provides an embodiment corresponding to a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps described in the above embodiment of the multi-domain collaborative sensing method for the power distribution Internet of Things.

[0136] It is understood that if the methods in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and executes all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0137] The computer-readable storage medium provided in this embodiment stores a computer program. When the processor executes the program, it can implement the following method: a multi-domain collaborative sensing method for power distribution IoT, which acquires raw data from the physical domain, information domain, energy domain, and social domain of the power distribution IoT; constructs a multi-domain heterogeneous information network based on the raw data; performs a meta-path-based random walk in the multi-domain heterogeneous information network to obtain homogeneous sequences among power distribution devices; discards irrelevant nodes other than power distribution devices in the homogeneous sequences to construct a homogeneous subgraph; trains a graph attention network model based on the homogeneous subgraph; and extracts the feature attributes of power distribution devices in the power distribution IoT through the trained graph attention network model. By establishing a multi-domain heterogeneous information network among the physical domain, information domain, energy domain, and social domain, and effectively sensing the endogenous relationships between devices through different meta-paths, and finally aggregating semantic information of different dimensions through the graph attention network model, the important feature attributes of devices are obtained, thereby improving the collaborative capability between interconnected power devices and solving the problem that traditional sensing methods rely too much on single-domain knowledge, ignore the massive multi-domain heterogeneous data in the power distribution IoT, and cannot guarantee the quality of differentiated device sensing data services.

[0138] The foregoing provides a detailed description of the multi-domain collaborative sensing method, apparatus, and medium for the power distribution Internet of Things (IoT) provided in this application. The various embodiments in the specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section. It should be noted that those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of the claims of this application.

[0139] It should also be noted that, in this specification, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A multi-domain collaborative sensing method for power distribution Internet of Things, characterized in that, include: Obtain raw data from the physical domain, information domain, energy domain, and social domain of the power distribution Internet of Things (IoT). Construct a multi-domain heterogeneous information network based on the original data; In the multi-domain heterogeneous information network, a meta-path-based random walk is performed to obtain homogeneous sequences among power distribution equipment; Discard irrelevant nodes in the homogeneous sequence except for the power distribution equipment, and construct a homogeneous subgraph; Train a graph attention network model based on the aforementioned homoprotic graph; The trained graph attention network model is used to extract the feature attributes of the power distribution equipment in the power distribution Internet of Things. The step of constructing a multi-domain heterogeneous information network based on the original data includes: Construct a first relationship graph between users based on the raw data of the social domain; A second relationship graph between users and power distribution equipment is constructed based on the raw data from the physical domain and the raw data from the social domain. Construct a third relationship graph between users and user information based on the original data of the social domain and the original data of the information domain; A fourth relationship diagram between power distribution equipment and power distribution equipment information is constructed based on the original data of the physical domain and the original data of the information domain. The first relationship diagram, the second relationship diagram, the third relationship diagram, and the fourth relationship diagram are integrated to obtain the multi-domain heterogeneous information network.

2. The multi-domain collaborative sensing method for power distribution IoT according to claim 1, characterized in that, The step of performing a meta-path-based random walk in the multi-domain heterogeneous information network to obtain homogeneous sequences among power distribution equipment includes: The initial meta-path between the power distribution equipment is determined according to preset screening conditions; The transition probability of each node is obtained based on the type, order, and weights between the nodes in the initial metapath; Starting from the initial node of the initial metapath, a random walk is performed at each new node according to the transition probability until a preset termination condition is reached; Store the homogeneous sequence generated by each meta-path walk.

3. The multi-domain collaborative sensing method for power distribution IoT according to claim 2, characterized in that, Determining the initial meta-path between the power distribution equipment based on preset filtering conditions includes: Obtain the average number of meta-path neighbors and the total number of nodes for each node; Determine whether the average number of the metapath neighbors of each node on each metapath divided by the total number of nodes is greater than a preset threshold; If so, then discard the currently described meta-path; If not, then record the current metapath as the initial metapath.

4. The multi-domain collaborative sensing method for power distribution IoT according to claim 1, characterized in that, The step of discarding irrelevant nodes (excluding power distribution equipment) in the homogeneous sequence and constructing a homogeneous subgraph includes: Discarding irrelevant nodes in the homogeneous sequence except for power distribution equipment, and linking nodes based on the same power distribution equipment, we obtain a homogeneous subgraph of the power distribution equipment; Based on the meta-path network in the homo-prime subgraph, the co-occurrence frequency of each edge in the homo-prime subgraph is calculated. Weights are assigned to edges in the homogeneous subgraph based on the co-occurrence frequency of each node.

5. The multi-domain collaborative sensing method for power distribution IoT according to claim 1, characterized in that, The step of extracting feature attributes of the power distribution equipment in the power distribution Internet of Things through the trained graph attention network model includes: The homoprotic graph is input into the trained graph attention network model for aggregation to obtain the feature attributes of each power distribution device in the power distribution Internet of Things.

6. The multi-domain collaborative sensing method for power distribution IoT according to claim 2, characterized in that, The preset termination condition is a preset number of steps.

7. A multi-domain collaborative sensing device for power distribution Internet of Things, characterized in that, include: The acquisition module is used to acquire raw data from the physical domain, information domain, energy domain, and social domain in the power distribution Internet of Things. The construction module is used to construct a multi-domain heterogeneous information network based on the original data; The sequence acquisition module is used to perform a meta-path-based random walk in the multi-domain heterogeneous information network to acquire homogeneous sequences among power distribution equipment. A homogeneous subgraph construction module is used to discard irrelevant nodes in the homogeneous sequence except for the power distribution equipment, and construct a homogeneous subgraph; The training model module is used to train a graph attention network model based on the homoprime subgraph; The analysis module is used to extract the feature attributes of the power distribution equipment in the power distribution Internet of Things through the trained graph attention network model; The step of constructing a multi-domain heterogeneous information network based on the original data includes: Construct a first relationship graph between users based on the raw data of the social domain; A second relationship graph between users and power distribution equipment is constructed based on the raw data from the physical domain and the raw data from the social domain. Construct a third relationship graph between users and user information based on the original data of the social domain and the original data of the information domain; A fourth relationship diagram between power distribution equipment and power distribution equipment information is constructed based on the original data of the physical domain and the original data of the information domain. The first relationship diagram, the second relationship diagram, the third relationship diagram, and the fourth relationship diagram are integrated to obtain the multi-domain heterogeneous information network.

8. A multi-domain collaborative sensing device for power distribution Internet of Things, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the multi-domain collaborative sensing method for the distribution Internet of Things as described in any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the multi-domain collaborative sensing method for the distribution Internet of Things as described in any one of claims 1 to 6.

Images