Method and device for determining equipment type of power distribution network equipment and storage medium

Abstract

The invention discloses a method and device for determining the equipment type of power distribution network equipment and a storage medium. The method comprises the steps that equipment information of power distribution network equipment is acquired, and the equipment information is used for indicating the performance of the power distribution network equipment; the equipment information is input into a target recognition model for analysis, a recognition result is obtained, and the target recognition model is obtained by training an initial recognition model of the power distribution network in response to the change of the equipment state and / or the connection relation of the power distribution network equipment; the initial recognition model is obtained by training a basic model based on a node feature vector of the power distribution network equipment in a topological structure; and determining the equipment type of the power distribution network equipment based on the identification result. The technical problem that the equipment identification efficiency of the power distribution network is low is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid equipment identification and classification, and in particular to a method, device and storage medium for determining the equipment type of power distribution network equipment. Background Art

[0002] As the scale and complexity of distribution networks increase, the main equipment and ancillary equipment of distribution networks are also gradually increasing. Accurately identifying the main equipment and ancillary equipment of distribution networks helps maintain the stability of power system operation. By timely discovering abnormal operation or aging of equipment, maintenance and replacement can be carried out in time to ensure the normal operation of equipment and avoid power outages or unstable power supply due to equipment failure. Therefore, accurate identification of main equipment (such as main transformers, lines, busbars, etc.) and ancillary equipment (such as switches, knife switches, etc.) is crucial to formulating reasonable maintenance plans, improving maintenance efficiency and ensuring power grid safety.

[0003] In the related technologies, traditional identification methods are difficult to cope with the needs of real-time dynamic changes, and there are problems such as high computational overhead and low identification efficiency in the operation and maintenance of the power grid. Therefore, there is a technical problem of low equipment recognition rate in the distribution network.

[0004] With respect to the above-mentioned technical problem of low equipment recognition rate in the distribution network, no effective solution has been proposed so far. Summary of the invention

[0005] The embodiments of the present invention provide a method, an apparatus and a storage medium for determining the device type of a distribution network device, so as to at least solve the technical problem of low efficiency in identifying devices in the distribution network.

[0006] According to one aspect of an embodiment of the present invention, a method for determining the device type of a distribution network device is provided. The method may include: obtaining device information of the distribution network device, wherein the device information is used to indicate the performance of the distribution network device; inputting the device information into a target recognition model for analysis to obtain a recognition result, wherein the target recognition model is obtained by training an initial recognition model of the distribution network in response to changes in the device state and / or connection relationship of the distribution network device, and the initial recognition model is obtained by training a basic model based on node feature vectors of the distribution network device in a topological structure; and determining the device type of the distribution network device based on the recognition result.

[0007] Optionally, the method for determining the device type of distribution network equipment also includes: in response to a change in the device state or connection relationship of the distribution network equipment, obtaining node information of a target device, wherein the target device is used to indicate a device whose device state or connection relationship has changed in the distribution network, and the node information is used to indicate information corresponding to the target device in a historical topology structure of the distribution network, and the historical topology structure is constructed using the historical device information of the distribution network equipment; based on the node information, obtaining a first node feature of an adjacent device, wherein the adjacent device is used to indicate a device of the distribution network adjacent to the target device in the historical topology structure, and the first node feature is used to indicate a position and / or function of the adjacent device in the historical topology structure; based on the node information of the target device and the node features of the adjacent devices, updating the historical topology structure to obtain a target topology structure; based on the target topology structure, training an initial recognition model of the distribution network to obtain a target recognition model.

[0008] Optionally, based on the node information of the target device and the node characteristics of the adjacent devices, the historical topology structure of the distribution network is updated to obtain the target topology structure, including: extracting a second node characteristic of the target device from the node information, wherein the second node characteristic is used to indicate the position and function of the target device in the historical topology structure; aggregating the first node characteristic with the second node characteristic to obtain an aggregated node characteristic; based on the aggregated node characteristic, updating the historical topology structure of the distribution network to obtain the target topology structure.

[0009] Optionally, based on the target topology, an initial identification model of the distribution network is trained to obtain a target identification model, including: obtaining basic status information of equipment in the distribution network in the target topology, wherein the basic status information is used to indicate the operating status of the equipment; based on the basic status information, determining target performance characteristics of the equipment, wherein the target performance characteristics are used to indicate at least one of the following information of the equipment in the target topology: location, connection relationship and performance; based on the target performance characteristics, an initial identification model of the distribution network is trained to obtain a target identification model.

[0010] Optionally, based on the target performance characteristics, an initial identification model of the distribution network is trained to obtain a target identification model, including: extracting historical performance characteristics of the equipment in the distribution network from the initial identification model, wherein the historical performance characteristics are used to indicate the position, connection relationship and performance of the equipment in the historical topology structure; generating comprehensive characteristics based on the historical performance characteristics and the target performance characteristics; based on the comprehensive characteristics, the initial identification model of the distribution network is trained to obtain a target identification model.

[0011] Optionally, the method for determining the equipment type of distribution network equipment also includes: obtaining multiple historical basic status information of the distribution network, wherein the historical basic status information is used to indicate the basic status information of the distribution network equipment before it is changed; based on each historical basic status information, constructing a historical topological structure of the distribution network; based on the historical topological structure, determining node feature vectors of multiple distribution network equipment, wherein the node feature vectors are used to indicate the attribute status of the distribution network equipment in the historical topological structure; based on each node feature vector, training a basic model to obtain an initial recognition model of the distribution network, wherein the basic model is used to indicate the status of each node in the historical topological structure.

[0012] According to another aspect of an embodiment of the present invention, a device for determining the device type of a distribution network device is also provided. The device may include: an acquisition unit, used to acquire device information of the distribution network device, wherein the device information is used to indicate the performance of the distribution network device; an identification unit, used to input the device information into a target identification model for analysis to obtain an identification result, wherein the target identification model is obtained by training an initial identification model of the distribution network in response to changes in the device status and / or connection relationship of the distribution network device, and the initial identification model is obtained by training a basic model based on node feature vectors of the distribution network device in a topological structure; and a determination unit, used to determine the device type of the distribution network device based on the identification result.

[0013] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored program, wherein when the program is executed by a processor, the device where the storage medium is located is controlled to execute the method for determining the device type of the distribution network device in the embodiment of the present invention.

[0014] According to another aspect of an embodiment of the present invention, a processor is further provided. The processor is used to run a program, wherein when the program is run, the method for determining the device type of the power distribution network device in the embodiment of the present invention is executed.

[0015] According to another aspect of the embodiments of the present invention, a computer program product is provided, which includes computer instructions, and when the computer instructions are executed by a processor, the method for determining the device type of the power distribution network device in the embodiments of the present invention is implemented.

[0016] In an embodiment of the present invention, the equipment information of the distribution network equipment is obtained; the equipment information is input into the target recognition model for analysis to obtain a recognition result, wherein the target recognition model is obtained by training the initial recognition model of the distribution network in response to changes in the equipment status and / or connection relationship of the distribution network equipment, and the initial recognition model is obtained by training the basic model based on the node feature vectors of the distribution network equipment in the topological structure; based on the recognition result, the equipment type of the distribution network equipment is determined. That is to say, the present invention obtains the target recognition model by training the initial recognition model of the distribution network, thereby identifying the equipment information of the distribution network equipment according to the target recognition model. Since the present invention trains the recognition model multiple times and updates it in real time, the technical problem of low equipment recognition efficiency of the distribution network is solved, and the technical effect of improving the equipment recognition efficiency of the distribution network is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0018] Figure 1 is a flow chart of a method for determining a device type of a distribution network device according to an embodiment of the present invention;

[0019] Figure 2 is a flow chart of a method for identifying power grid equipment based on functional hierarchy division and graph embedding technology according to an embodiment of the present invention;

[0020] Figure 3 4 is a schematic diagram of a device for determining a device type of a distribution network device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0022] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, functional component or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, functional components or devices.

[0023] According to an embodiment of the present invention, an embodiment of a method for determining the device type of a distribution network device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0024] Figure 1 is a flow chart of a method for determining a device type of a distribution network device according to an embodiment of the present invention. Figure 1 As shown, the method may include the following steps:

[0025] Step S101, obtaining equipment information of distribution network equipment.

[0026] In the technical solution provided in the above step S101 of the present invention, the device information is used to indicate the performance of the distribution network equipment.

[0027] In this embodiment, the device information of the distribution network equipment is obtained, for example, by using a sensor or a camera, which is only an example and does not limit the specific method of obtaining the device information of the distribution network equipment.

[0028] Step S102, input the device information into the target recognition model for analysis to obtain a recognition result.

[0029] In the technical solution provided in the above step S102 of the present invention, the target recognition model is obtained by training the initial recognition model of the distribution network in response to changes in the device state and / or connection relationship of the distribution network equipment, and the initial recognition model is obtained by training the basic model based on the node feature vector of the distribution network equipment in the topological structure. Among them, the target recognition model can also be called the final model for identifying the primary and secondary stations.

[0030] In this embodiment, after acquiring the device information of the distribution network device in step S101, the device information is input into the target recognition model for analysis to obtain a recognition result.

[0031] Optionally, when the device state or connection relationship of the distribution network device changes, the node information of the target device is obtained, wherein the target device is used to indicate the device whose device state or connection relationship has changed in the distribution network, and the node information is used to indicate the corresponding information of the target device in the historical topology structure of the distribution network, and the historical topology structure is constructed using the historical device information of the distribution network device. The target device can also become an affected node (denoted as v).

[0032] For example, real-time monitoring of changes in device status and connection relationships in the power grid, detection of updates to the local topology, and acquisition of node information corresponding to new devices or device status changes when new devices are added or device status changes are made.

[0033] Optionally, after acquiring the node information, the first node feature of the adjacent device is acquired, wherein the adjacent device is used to indicate the device of the distribution network adjacent to the target device in the historical topology structure, and the first node feature is used to indicate the location and / or function of the adjacent device in the historical topology structure. The first node feature can also be referred to as a feature after the neighbor node is aggregated.

[0034] For example, for the affected node v, a fixed number S of nodes are randomly resampled from its new neighbor node set N(v). The specific collection method can be expressed by the following formula (1):

[0035]

[0036] Optionally, the historical topology structure is updated according to the node information of the target device and the node characteristics of the adjacent devices to obtain the target topology structure.

[0037] For example, a second node feature of the target device is extracted from the node information, wherein the second node feature is used to indicate the position and function of the target device in the historical topology structure; the first node feature and the second node feature are aggregated to obtain an aggregated node feature; and based on the aggregated node feature, the historical topology structure of the distribution network is updated to obtain a target topology structure.

[0038] For another example, the features of the sampled neighbor nodes are subjected to mean aggregation with low computational complexity. The specific aggregation method can be calculated by the following formula (2):

[0039]

[0040] in, It can represent the feature vector of the neighbor node u at the k-1th layer. The second node feature is combined with the features of the neighbor nodes of the first node feature after aggregation, and the node embedding vector is updated. The specific update method can be calculated by the following formula (3):

[0041]

[0042] Among them, σ can represent the activation function, W (k) can represent the weight matrix of the kth layer, CONCAT can represent the vector connection operation, and through multi-layer aggregation and updating, a low-dimensional embedding vector of the node is finally generated, that is, the final embedding vector. The specific generation method can be calculated by the following formula (4):

[0043]

[0044] Among them, K can represent the number of layers of the GraphSAGE (Graph Sampling and Aggregation) model, z v The final embedding vector of node v can be represented, so that the historical topology is updated according to the final embedding vector to obtain the target topology.

[0045] Optionally, according to the target topology, an initial recognition model of the distribution network is trained to obtain a target recognition model.

[0046] For example, basic status information of equipment in the distribution network in the target topology is obtained, wherein the basic status information is used to indicate the operating status of the equipment; based on the basic status information, the target performance characteristics of the equipment are determined, wherein the target performance characteristics are used to indicate at least one of the following information of the equipment in the target topology: location, connection relationship and performance.

[0047] For another example, the voltage, current, and switch state of the device over a period of time are obtained, and then the statistical feature quantities of the mean, variance, maximum value, and minimum value are calculated to generate the target performance characteristics, that is, the time series characteristics t v , specifically generate time series features t v The method is shown in the following formula (5):

[0048] t v =[mean(t v ), var(t v ), max(t v ), min(t v )] (5)

[0049] Optionally, historical performance features of the devices in the distribution network are extracted from the initial recognition model, wherein the historical performance features are used to indicate the location, connection relationship and performance of the devices in the historical topology. The historical performance features can also be called graph embedding features. v For example, the graph embedding feature z can be calculated by the above formula (4): v , which will not be elaborated here.

[0050] Optionally, a comprehensive feature is generated based on the historical performance feature and the target performance feature; based on the comprehensive feature, an initial recognition model of the distribution network is trained to obtain a target recognition model.

[0051] For example, when generating the time series t v After that, the graph is embedded into feature z v With time series characteristics t v Normalized and then connected together to form a comprehensive feature vector f v , specifically forming a comprehensive feature vector f v The method is shown in the following formula (6):

[0052] f v =CONCAT(z v , t v ) (6)

[0053] Among them, CONCAT can represent a vector concatenation operation.

[0054] Step S103: determining the device type of the distribution network device based on the identification result.

[0055] In the technical solution provided in the above step S103 of the present invention, the equipment types of the distribution network equipment can at least be: main equipment (such as main transformer, line, busbar, etc.) and auxiliary equipment (such as switch, knife switch, etc.).

[0056] In this embodiment, after obtaining the identification result in step S102, the device type of the power distribution network device is determined according to the identification result.

[0057] Optionally, after determining the device type, the device can be managed and maintained more accurately, improving management efficiency.

[0058] It should be noted that the above embodiment can be executed by a device for determining the device type of distribution network equipment.

[0059] In the above steps S101 to S103 of the present invention, the equipment information of the distribution network equipment is obtained; the equipment information is input into the target recognition model for analysis to obtain the recognition result, wherein the target recognition model is obtained by training the initial recognition model of the distribution network in response to the change of the equipment status and / or connection relationship of the distribution network equipment, and the initial recognition model is obtained by training the basic model based on the node feature vector of the distribution network equipment in the topological structure; based on the recognition result, the equipment type of the distribution network equipment is determined. That is to say, the present invention obtains the target recognition model by training the initial recognition model of the distribution network, thereby identifying the equipment information of the distribution network equipment according to the target recognition model. Since the present invention trains the recognition model multiple times and updates it in real time, the technical problem of low equipment recognition efficiency of the distribution network is solved, and the technical effect of improving the equipment recognition efficiency of the distribution network is achieved.

[0060] The above method of this embodiment is further introduced below.

[0061] As an optional implementation manner, the method for determining the device type of distribution network equipment also includes: in response to a change in the device state or connection relationship of the distribution network equipment, obtaining node information of a target device, wherein the target device is used to indicate a device whose device state or connection relationship has changed in the distribution network, and the node information is used to indicate the corresponding information of the target device in the historical topology structure of the distribution network, and the historical topology structure is constructed using the historical device information of the distribution network equipment; based on the node information, obtaining a first node feature of an adjacent device, wherein the adjacent device is used to indicate a device of the distribution network adjacent to the target device in the historical topology structure, and the first node feature is used to indicate the position and / or function of the adjacent device in the historical topology structure; based on the node information of the target device and the node features of the adjacent devices, updating the historical topology structure to obtain a target topology structure; based on the target topology structure, training an initial recognition model of the distribution network to obtain a target recognition model.

[0062] In this embodiment, when the device state or connection relationship of the distribution network device changes, the node information of the target device and the first node feature of the adjacent device are obtained, and the historical topology structure is updated according to the node information of the target device and the node features of the adjacent device to obtain the target topology structure; thus, the initial recognition model of the distribution network is trained according to the target topology structure to obtain the target recognition model. Among them, the target device can also be an affected node (denoted as v), and the first node feature can also be called the feature after the neighbor node is aggregated.

[0063] Optionally, the node information of the target device and the first node characteristics of the adjacent device are obtained as shown in the above formula (1), which will not be described again here.

[0064] Optionally, by timely updating the topology structure, the device status and connection relationship in the distribution network can be more accurately identified, thereby more effectively monitoring and managing the operating status of the distribution network and improving operating efficiency.

[0065] As an optional implementation method, based on the node information of the target device and the node characteristics of the adjacent devices, the historical topology structure of the distribution network is updated to obtain the target topology structure, including: extracting the second node characteristic of the target device from the node information, wherein the second node characteristic is used to indicate the position and function of the target device in the historical topology structure; aggregating the first node characteristic with the second node characteristic to obtain the aggregated node characteristic; based on the aggregated node characteristic, updating the historical topology structure of the distribution network to obtain the target topology structure.

[0066] In this embodiment, a second node feature of the target device is extracted from the node information, wherein the second node feature is used to indicate the location and function of the target device in the historical topology structure; the first node feature and the second node feature are aggregated to obtain an aggregate node feature; based on the aggregate node feature, the historical topology structure of the distribution network is updated to obtain a target topology structure. The second node feature may also be referred to as a feature of a neighboring node.

[0067] Optionally, a specific method of aggregating the first node feature and the second node feature to obtain the aggregate node feature is as shown in the above formula (2) and formula (3), which will not be repeated here.

[0068] Optionally, by aggregating the first node feature with the second node feature, the relationship between the target device and its surrounding neighbor nodes can be considered more comprehensively, thereby better understanding the characteristics and properties of the entire topology. Updating the historical topology of the distribution network can help the system better adapt and respond to changing environments and demands, and improve the robustness and reliability of the system.

[0069] As an optional implementation method, based on the target topology structure, an initial identification model of the distribution network is trained to obtain a target identification model, including: obtaining basic status information of the equipment of the distribution network in the target topology structure, wherein the basic status information is used to indicate the operating status of the equipment; based on the basic status information, determining the target performance characteristics of the equipment, wherein the target performance characteristics are used to indicate at least one of the following information of the equipment in the target topology structure: location, connection relationship and performance; based on the target performance characteristics, the initial identification model of the distribution network is trained to obtain a target identification model.

[0070] In this embodiment, basic state information of the equipment in the distribution network in the target topology is obtained; the target performance characteristics of the equipment are determined based on the basic state information, and then the initial recognition model of the distribution network is trained based on the target performance characteristics to obtain the target recognition model. The target performance characteristics can also be called time series characteristics t v .

[0071] Optionally, for the time series feature t v The calculation of is shown in the above formula (5), which will not be repeated here.

[0072] Optionally, based on the target performance characteristics, the initial recognition model of the distribution network can be trained to improve the accuracy and reliability of the model. The target recognition model obtained through training can more accurately identify the equipment in the distribution network and accurately determine the status and operation of the equipment.

[0073] As an optional implementation method, based on the target performance characteristics, the initial identification model of the distribution network is trained to obtain the target identification model, including: extracting the historical performance characteristics of the equipment in the distribution network from the initial identification model, wherein the historical performance characteristics are used to indicate the position, connection relationship and performance of the equipment in the historical topology structure; generating comprehensive characteristics based on the historical performance characteristics and the target performance characteristics; based on the comprehensive characteristics, the initial identification model of the distribution network is trained to obtain the target identification model.

[0074] In this embodiment, the historical performance characteristics of the equipment in the distribution network are extracted from the initial recognition model; the comprehensive characteristics are generated based on the historical performance characteristics and the target performance characteristics; the initial recognition model of the distribution network is trained based on the comprehensive characteristics to obtain the target recognition model. The historical performance characteristics can also be called graph embedding features. v .

[0075] Optionally, the graph embedding feature z can be calculated by the above formula (4): v , the comprehensive characteristics can be calculated by the above formula (6), which will not be repeated here.

[0076] Optionally, the accuracy and reliability of the model can be improved by training the initial recognition model: by training with comprehensive historical performance characteristics and target performance characteristics, the performance characteristics of the equipment can be better captured, improving the accuracy and reliability of the model.

[0077] As an optional implementation method, the method for determining the equipment type of distribution network equipment also includes: obtaining multiple historical basic status information of the distribution network, wherein the historical basic status information is used to indicate the basic status information of the distribution network equipment before it is changed; based on each historical basic status information, constructing a historical topological structure of the distribution network; based on the historical topological structure, determining node feature vectors of multiple distribution network equipment, wherein the node feature vectors are used to indicate the attribute status of the distribution network equipment in the historical topological structure; based on each node feature vector, training a basic model to obtain an initial recognition model of the distribution network, wherein the basic model is used to indicate the status of each node in the historical topological structure.

[0078] In this embodiment, multiple historical basic state information of the distribution network is obtained, and based on each historical basic state information, a historical topological structure of the distribution network is constructed; based on the historical topological structure, node feature vectors of multiple distribution network devices are determined, and based on each node feature vector, a basic model is trained to obtain an initial recognition model of the distribution network. The node feature vector can also be called a topological attribute, and the topological attribute can include node degree, degree centrality, and betweenness centrality.

[0079] For example, by collecting the connection relationship data between all devices in the distribution network, constructing the topology structure and generating the node vector, the node degree can be calculated according to formula (7):

[0080]

[0081] Among them, d i It can represent the degree of node i, A ij It can represent the elements of the adjacency matrix. Degree Centrality can be calculated according to the following formula (8):

[0082]

[0083] Among them, C D (i) is the degree centrality of node i, and N is the total number of nodes. The betweenness centrality can be calculated according to the following formula (9):

[0084]

[0085] Among them, C B (i) is the betweenness centrality of node i, σ is the number of shortest paths from node s to node t, σ st (i) is the number of nodes i passed through in the shortest path from node s to node t.

[0086] It should be noted that the above embodiment can be executed by a device for determining the device type of distribution network equipment.

[0087] In this embodiment, the equipment information of the distribution network equipment is obtained; the equipment information is input into the target recognition model for analysis to obtain the recognition result, wherein the target recognition model is obtained by training the initial recognition model of the distribution network in response to the change of the equipment status and / or connection relationship of the distribution network equipment, and the initial recognition model is obtained by training the basic model based on the node feature vector of the distribution network equipment in the topological structure; based on the recognition result, the equipment type of the distribution network equipment is determined. That is to say, the present invention obtains the target recognition model by training the initial recognition model of the distribution network, thereby identifying the equipment information of the distribution network equipment according to the target recognition model. Since the present invention trains the recognition model multiple times and updates it in real time, the technical problem of low efficiency of equipment recognition of the distribution network is solved, and the technical effect of improving the efficiency of equipment recognition of the distribution network is achieved.

[0088] The technical solution of the embodiment of the present invention is illustrated below in conjunction with preferred implementation modes.

[0089] As the scale and complexity of distribution networks increase, the main equipment and ancillary equipment of distribution networks are also gradually increasing. Accurately identifying the main equipment and ancillary equipment of distribution networks helps maintain the stability of power system operation. By timely discovering abnormal operation or aging of equipment, maintenance and replacement can be carried out in time to ensure the normal operation of equipment and avoid power outages or unstable power supply due to equipment failure. Therefore, accurate identification of main equipment (such as main transformers, lines, busbars, etc.) and ancillary equipment (such as switches, knife switches, etc.) is crucial to formulating reasonable maintenance plans, improving maintenance efficiency and ensuring power grid safety.

[0090] In the related technologies, traditional identification methods are difficult to cope with the needs of real-time dynamic changes, and there are problems such as high computational overhead and low identification efficiency. In the operation and maintenance of the power grid, there is a technical problem of low equipment identification rate in the distribution network. There is currently no effective solution to the above-mentioned technical problem of low equipment identification efficiency in the distribution network.

[0091] However, an embodiment of the present invention proposes a method for identifying power grid equipment based on functional hierarchy division and graph embedding technology, which establishes a topological structure by collecting the connection relationship and equipment status information of the distribution network equipment, and establishes a graph embedding model based on the nodes in the topological structure, monitors the changes in the equipment status and connection relationship in the power grid in real time, and detects the update of the local topological structure. When there is a new device or the equipment status changes in the topological structure, the graph embedding features in the graph embedding model are normalized with the updated node features, the topological structure is updated again, the recognition model is trained, and the main and auxiliary equipment of the distribution network are identified according to the trained recognition model, which solves the technical problem of low efficiency in identifying equipment in the distribution network and achieves the technical effect of improving the efficiency of identifying equipment in the distribution network.

[0092] The embodiments of the present invention are further described below.

[0093] Figure 2 : is a flowchart of a method for identifying power grid equipment based on functional hierarchy division and graph embedding technology according to an embodiment of the present invention. The analysis method comprises the following steps:

[0094] Step S201: topology construction and node vector generation.

[0095] In this embodiment, by collecting connection relationship data between all devices in the power distribution network, etc., a topological structure is constructed and a node vector is generated.

[0096] Optionally, first collect data on the distribution network: collect the connection relationship data between all devices in the distribution network, and build a relationship graph between nodes (devices) and edges (connections). Collect the attribute data of the equipment, including equipment type, such as transformer, line, busbar, switch, knife switch, etc., and installation location, rated power, voltage level, etc. Then collect the voltage, current, and switch status data of the equipment from the SCADA (Supervisory Control and Data Acquisition) system, sensors or monitoring equipment.

[0097] Optionally, topology construction is performed based on the collected data: the data is standardized, and the collected device connection relationship data is used to construct a global topology map of the power grid. The topology of the power grid is represented by an adjacency matrix or adjacency table, where the nodes in the graph represent devices and the edges represent the connection relationships between devices.

[0098] Optionally, after the topology is constructed, the topological properties are calculated using the topological structure, wherein the topological properties may include node degree, degree centrality, and betweenness centrality. The node degree may be calculated according to the aforementioned formula (7), the degree centrality may be calculated according to the aforementioned formula (8), and the betweenness centrality may be calculated according to the aforementioned formula (9), which will not be described in detail here.

[0099] Optionally, a node feature vector is generated based on the calculated attribute data: a feature vector of each node is generated based on the attribute data, state data and topological attributes of the device for subsequent graph embedding model training.

[0100] Step S202: functional level division.

[0101] In this embodiment, after the node feature vector is generated in step S201, the devices are divided into functional levels.

[0102] Optionally, equipment classification: Based on the functions and characteristics of the power grid equipment, the equipment classification is defined as follows: Generation layer: including generators, main transformers and other equipment, responsible for power generation and voltage boosting. Transmission layer: including high-voltage transmission lines, busbars in substations, transformers and other equipment, responsible for long-distance power transmission. Distribution layer: including distribution transformers, distribution lines, distribution cabinets and other equipment, responsible for distributing electricity to end users. End user layer: including meters and load equipment on the user side, responsible for the final consumption of electricity.

[0103] Optionally, the devices are assigned to corresponding levels according to their attribute data and classification.

[0104] Optionally, hierarchical division and attribute labeling: further refine hierarchical division according to the device connection relationship. Use the graph algorithm of the minimum spanning tree (MST) to construct the minimum spanning tree and perform hierarchical division to ensure that the devices at the same level have a close connection relationship. After the hierarchical division is completed, the corresponding hierarchical attributes are marked for the devices at each level, that is, each device is marked with its own level, and each level is numbered for subsequent device identification and management.

[0105] For example, the power generation layer is numbered 1, the transmission layer is numbered 2, the distribution layer is numbered 3, and the end-user layer is numbered 4; the generator and the main transformer are marked as the power generation layer, and the transmission line and transformer are marked as the transmission layer.

[0106] Step S203: establish and train a graph embedding model.

[0107] In this embodiment, a graph embedding model is established and trained based on the data collected in step S201 and the classification result of step S202.

[0108] Optionally, the GraphSAGE algorithm is used to build the graph embedding model. GraphSAGE can efficiently process large-scale graph data and is suitable for the application scenario of power grid equipment. GraphSAGE can process dynamic graphs of power grid topology and support incremental updates by sampling and aggregating information of neighboring nodes without retraining the entire graph. Its implementation method is as follows:

[0109] Optionally, feature aggregation: for each node v, randomly sample a fixed number S of nodes from its neighbor node set N(v), perform mean aggregation on the features of the sampled neighbor nodes with low computational complexity, combine the node's own features with the aggregated features of the neighbor nodes, update the node embedding vector, and finally generate a low-dimensional embedding vector of the node through multi-layer aggregation and updating, and select a suitable loss function and optimizer for gradient descent optimization. Update the model parameters through multiple iterations, minimize the loss function, and obtain the final embedding vector of each node.

[0110] Optionally, the specific collection method can be calculated by the aforementioned formula (1), the specific aggregation method can be calculated by the aforementioned formula (2), the specific update method can be calculated by the aforementioned formula (3), and the specific generation method can be calculated by the aforementioned formula (4), which will not be repeated here.

[0111] Optionally, for each node v, a fixed number S of nodes are randomly sampled from its neighbor node set N(v). The specific collection method can be expressed by the aforementioned formula (1), which will not be described in detail here.

[0112] Optionally, the features of the sampled neighbor nodes are subjected to mean aggregation with low computational complexity. The specific aggregation method can be calculated by the aforementioned formula (2), which will not be described in detail here.

[0113] Optionally, node embedding update: combine the node's own features with the aggregated features of neighboring nodes to update the node embedding vector. The specific updating method can be calculated by the aforementioned formula (3), which will not be described in detail here.

[0114] Optionally, multi-layer embedding, through multi-layer aggregation and updating, finally generates a low-dimensional embedding vector of the node. The specific generation method can be calculated by the aforementioned formula (4), which will not be repeated here.

[0115] Step S204: real-time local topology update and fast embedding calculation.

[0116] In this embodiment, the changes in the device status and connection relationship in the power grid are monitored in real time, the update of the local topology structure is detected, and when a new device is added or the device status changes, the embedding vector of the affected node is recalculated. Here, only the embedding vectors of the affected node and its neighboring nodes are updated to avoid retraining the entire model and improve computing efficiency.

[0117] Optionally, for node resampling and local feature reaggregation, for the affected node v, a fixed number S of nodes are randomly resampled from its new neighbor node set N(v). The features of the resampled neighbor nodes are aggregated, and mean aggregation is still used. The specific aggregation method can be calculated by the above formula (2), which will not be repeated here.

[0118] Optionally, for node embedding re-update, the node's own features are combined with the re-aggregated neighbor node features to update the node embedding vector. The specific updating method can be calculated by the aforementioned formula (3), which will not be described in detail here.

[0119] Step S205, comprehensive feature fusion and primary and secondary station classification.

[0120] In this embodiment, comprehensive feature fusion is calculated based on the graph embedding model and the updated node vector, and primary and secondary station classification is implemented.

[0121] Optionally, record the voltage, current, and switch state of the device over a period of time, and then calculate the statistical feature quantities of the mean, variance, maximum, and minimum to generate the time series feature t v , specifically generate time series features t v The method of is shown in the above formula (5), which will not be described again here.

[0122] Optionally, when generating the time series t v After that, the graph is embedded into feature z v With the time series feature f v Normalized and then connected together to form a comprehensive feature vector f v , specifically forming a comprehensive feature vector f v The method of is shown in the above formula (6), which will not be described again here.

[0123] Optionally, when generating the comprehensive feature vector f v After that, the comprehensive feature vector f of all devices v and its corresponding labels are used as the training data set for the online random forest (ORF) classification model. After certain verification and optimization, the final model for identifying the primary and secondary stations is obtained.

[0124] Optionally, online random forest is an ensemble learning algorithm that supports incremental learning. When new data arrives, only the affected part of the trees and leaf nodes are incrementally updated, and finally the affected part of the trees is retrained using the updated data.

[0125] Optionally, the present invention combines the graph-embedded GraphSAGE algorithm, time series feature fusion, and online random forest classification model to support real-time incremental updates, achieving rapid updates and efficient identification when the topology changes. By generating graph embedding features and combining them with time series features to form a comprehensive feature vector, the efficiency, accuracy, and reliability of device identification are significantly improved. At the same time, the functional hierarchy division and dynamic topology detection methods enable the present invention to efficiently identify the main equipment and auxiliary equipment in the power grid, and quickly analyze the initial state when the equipment state changes, and construct an equipment ticket model. Compared with the prior art, the present invention exhibits significant advantages in efficient identification, dynamic adaptability, scalability, real-time update capability, and low computational overhead, overcomes the shortcomings of the prior art, and improves the intelligence level and reliability of power grid equipment management.

[0126] In this embodiment, the present invention proposes a method for identifying power grid equipment based on functional hierarchy division and graph embedding technology, which realizes efficient, real-time and accurate identification of main equipment (such as main transformer, line, busbar, etc.) and auxiliary equipment (such as switch, knife switch, etc.) in power grid equipment through functional hierarchy division, power grid topology structure analysis, graph embedding technology and time series feature extraction. The present invention can quickly update the device embedding vector and perform identification when the power grid topology structure undergoes local changes, and is suitable for dynamic management and maintenance of power grids.

[0127] In this embodiment, a topological structure is established by collecting connection relationships and equipment status information of distribution network equipment, and a graph embedding model is established based on the nodes in the topological structure. Changes in equipment status and connection relationships in the power grid are monitored in real time, and updates to the local topological structure are detected. When new equipment is added to the topological structure or equipment status changes, the graph embedding features in the graph embedding model are normalized with the updated node features, the topological structure is updated again, the recognition model is trained, and the primary and secondary equipment of the distribution network are identified based on the trained recognition model, thereby solving the technical problem of low equipment recognition efficiency in the distribution network and achieving the technical effect of improving equipment recognition efficiency in the distribution network.

[0128] According to an embodiment of the present invention, a device for determining the device type of a distribution network device is also provided. It should be noted that the device for determining the device type of a distribution network device can be used to execute the method for determining the device type of a distribution network device in Embodiment 1.

[0129] Figure 3 Schematic diagram of a device for determining the device type of a distribution network device according to an embodiment of the present invention. Figure 3 As shown, the device 300 for determining the device type of the power distribution network device may include: an acquisition unit 301 , an identification unit 302 and a determination unit 303 .

[0130] The acquisition unit 301 is used to acquire device information of a distribution network device, wherein the device information is used to indicate the performance of the distribution network device.

[0131] The identification unit 302 is used to input the device information into the target identification model for analysis to obtain an identification result, wherein the target identification model is obtained by training the initial identification model of the distribution network in response to changes in the device status and / or connection relationship of the distribution network equipment, and the initial identification model is obtained by training the basic model based on the node feature vectors of the distribution network equipment in the topological structure.

[0132] The determination unit 303 is used to determine the device type of the distribution network device based on the identification result.

[0133] Optionally, the device 300 for determining the device type of the distribution network device may further include: a first acquisition unit, used to acquire node information of a target device in response to a change in the device state or connection relationship of the distribution network device, wherein the target device is used to indicate a device whose device state or connection relationship has changed in the distribution network, and the node information is used to indicate information corresponding to the target device in a historical topology structure of the distribution network, and the historical topology structure is constructed using the historical device information of the distribution network device; a second acquisition unit, used to acquire a first node feature of an adjacent device based on the node information, wherein the adjacent device is used to indicate a device of the distribution network adjacent to the target device in the historical topology structure, and the first node feature is used to indicate a position and / or function of the adjacent device in the historical topology structure; an updating unit, used to update the historical topology structure based on the node information of the target device and the node features of the adjacent devices to obtain a target topology structure; a first training unit, used to train an initial recognition model of the distribution network based on the target topology structure to obtain a target recognition model.

[0134] Optionally, the update unit may include: an extraction module, used to extract a second node feature of the target device from the node information, wherein the second node feature is used to indicate the position and function of the target device in the historical topology structure; an aggregation module, used to aggregate the first node feature with the second node feature to obtain an aggregated node feature; and an update module, used to update the historical topology structure of the distribution network based on the aggregated node feature to obtain a target topology structure.

[0135] Optionally, the training unit may include: an acquisition module for acquiring basic status information of equipment in the distribution network in the target topology, wherein the basic status information is used to indicate the operating status of the equipment; a determination module for determining target performance characteristics of the equipment based on the basic status information, wherein the target performance characteristics are used to indicate at least one of the following information of the equipment in the target topology: location, connection relationship and performance; a training module for training an initial identification model of the distribution network based on the target performance characteristics to obtain a target identification model.

[0136] Optionally, the training module may include: an extraction submodule for extracting historical performance characteristics of equipment in the distribution network from the initial identification model, wherein the historical performance characteristics are used to indicate the location, connection relationship and performance of the equipment in the historical topology structure; a generation submodule for generating comprehensive characteristics based on the historical performance characteristics and the target performance characteristics; and a training submodule for training the initial identification model of the distribution network based on the comprehensive characteristics to obtain a target identification model.

[0137] Optionally, the device 300 for determining the equipment type of distribution network equipment may further include: a third acquisition unit, used to acquire multiple historical basic status information of the distribution network, wherein the historical basic status information is used to indicate the basic status information of the distribution network equipment before it is changed; a construction unit, used to construct a historical topological structure of the distribution network based on each historical basic status information; a first determination unit, used to determine node feature vectors of multiple distribution network equipment based on the historical topological structure, wherein the node feature vectors are used to indicate the attribute status of the distribution network equipment in the historical topological structure; a second training unit, used to train the basic model based on each node feature vector to obtain an initial recognition model of the distribution network, wherein the basic model is used to indicate the status of each node in the historical topological structure.

[0138] In this embodiment, the equipment information of the distribution network equipment is obtained; the equipment information is input into the target recognition model for analysis to obtain the recognition result, wherein the target recognition model is obtained by training the initial recognition model of the distribution network in response to the change of the equipment status and / or connection relationship of the distribution network equipment, and the initial recognition model is obtained by training the basic model based on the node feature vector of the distribution network equipment in the topological structure; based on the recognition result, the equipment type of the distribution network equipment is determined. That is to say, the present invention obtains the target recognition model by training the initial recognition model of the distribution network, thereby identifying the equipment information of the distribution network equipment according to the target recognition model. Since the present invention trains the recognition model multiple times and updates it in real time, the technical problem of low efficiency of equipment recognition of the distribution network is solved, and the technical effect of improving the efficiency of equipment recognition of the distribution network is achieved.

[0139] According to an embodiment of the present invention, a computer-readable storage medium is further provided. The storage medium includes a stored program, wherein the program executes the method for determining the device type of the power distribution network device in Embodiment 1.

[0140] According to an embodiment of the present invention, a processor is further provided, the processor being used to run a program, wherein the method for determining the device type of the distribution network device in Embodiment 1 is executed when the program is run.

[0141] According to an embodiment of the present invention, a computer program product is further provided. The computer program product includes computer instructions. When the computer instructions are executed by a processor, the method for determining the device type of the distribution network device in Embodiment 1 is implemented.

[0142] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0143] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0144] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0145] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0146] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0147] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent functional component, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, 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 functional component, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc., which can store program codes.

[0148] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for determining the equipment type of a distribution network equipment, characterized in that: include: Acquire device information of the distribution network device, wherein the device information is used to indicate the performance of the distribution network device; The device information is input into a target recognition model for analysis to obtain a recognition result, wherein the target recognition model is obtained by training an initial recognition model of the distribution network in response to changes in the device state and / or connection relationship of the distribution network device, and the initial recognition model is obtained by training a basic model based on node feature vectors of the distribution network device in a topological structure; Based on the identification result, the device type of the power distribution network device is determined.

2. The method according to claim 1, characterized in that The method further comprises: In response to a change in the device state or the connection relationship of the distribution network device, acquiring node information of a target device, wherein the target device is used to indicate a device in the distribution network whose device state or the connection relationship has changed, and the node information is used to indicate information corresponding to the target device in a historical topology structure of the distribution network, wherein the historical topology structure is constructed using historical device information of the distribution network device; Based on the node information, obtaining a first node feature of an adjacent device, wherein the adjacent device is used to indicate the device of the distribution network adjacent to the target device in the historical topology structure, and the first node feature is used to indicate the position and / or function of the adjacent device in the historical topology structure; Based on the node information of the target device and the node characteristics of the adjacent devices, the historical topology structure is updated to obtain a target topology structure; Based on the target topology, the initial recognition model of the distribution network is trained to obtain the target recognition model.

3. The method according to claim 2, characterized in that Based on the node information of the target device and the node characteristics of the adjacent device, updating the historical topology structure of the distribution network to obtain a target topology structure includes: Extracting a second node feature of the target device from the node information, wherein the second node feature is used to indicate a position and a function of the target device in the historical topology structure; Aggregating the first node feature and the second node feature to obtain an aggregated node feature; Based on the aggregation node characteristics, the historical topology of the distribution network is updated to obtain the target topology.

4. The method according to claim 2, characterized in that: Based on the target topology structure, the initial recognition model of the distribution network is trained to obtain the target recognition model, including: Acquire basic status information of the device of the distribution network in the target topology structure, wherein the basic status information is used to indicate the operating status of the device; Determine a target performance characteristic of the device based on the basic status information, wherein the target performance characteristic is used to indicate at least one of the following information of the device in the target topology: location, connection relationship, and performance; Based on the target performance characteristics, the initial recognition model of the distribution network is trained to obtain the target recognition model.

5. The method according to claim 4, characterized in that Based on the target performance characteristics, the initial recognition model of the distribution network is trained to obtain the target recognition model, including: Extracting historical performance characteristics of the device of the distribution network from the initial identification model, wherein the historical performance characteristics are used to indicate the location, connection relationship and performance of the device in the historical topology; generating a comprehensive feature based on the historical performance feature and the target performance feature; Based on the comprehensive features, the initial recognition model of the distribution network is trained to obtain the target recognition model.

6. The method according to any one of claims 1 to 5, characterized in that The method further comprises: Acquire multiple historical basic status information of the distribution network, wherein the historical basic status information is used to indicate the basic status information of the distribution network equipment before it is changed; Based on each of the historical basic status information, construct the historical topological structure of the distribution network; Based on the historical topological structure, determining node feature vectors of a plurality of the distribution network devices, wherein the node feature vectors are used to indicate attribute states of the distribution network devices in the historical topological structure; The basic model is trained based on the feature vectors of each node to obtain the initial identification model of the distribution network, wherein the basic model is used to indicate the status of each node in the historical topology structure.

7. A device for determining the equipment type of a distribution network equipment, characterized in that: include: An acquiring unit, configured to acquire device information of the distribution network device, wherein the device information is used to indicate the performance of the distribution network device; An identification unit, used for inputting the device information into a target identification model for analysis to obtain an identification result, wherein the target identification model is obtained by training an initial identification model of the distribution network in response to a change in the device state and / or connection relationship of the distribution network device, and the initial identification model is obtained by training a basic model based on a node feature vector of the distribution network device in a topological structure; A determination unit is used to determine the device type of the distribution network device based on the identification result.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed by a processor, the device where the storage medium is located is controlled to execute the method according to any one of claims 1 to 6.

9. A processor, characterized in that: The processor is used to run a program, wherein the program executes the method according to any one of claims 1 to 6 when running.

10. A computer program product, characterized in that The computer program product comprises computer instructions, which implement the method according to any one of claims 1 to 6 when executed by a processor.

Images