Generation method and device of fault inspection strategy of power distribution network and storage medium

Abstract

The invention discloses a power distribution network fault inspection strategy generation method and device and a storage medium. The method comprises the following steps: under the condition that the power distribution network is in a fault, acquiring single line diagram data and geographic position data of the power distribution network and initial state data of a plurality of hardware devices in the power distribution network; determining an initial topological structure of the power distribution network based on the single line diagram data and the geographic position data; inputting the initial topological structure and the initial state data into a target topological model of the power distribution network for analysis to obtain to-be-patrolled areas and to-be-patrolled devices of the plurality of power distribution networks; based on the target inspection resource information of the power distribution network, the target inspection constraint condition of the power distribution network and the target inspection function of the power distribution network, inspecting the to-be-inspected area and the to-be-inspected equipment to obtain an inspection result; and generating a fault patrol strategy of the power distribution network based on the patrol result. The technical problem of low fault inspection efficiency of the power distribution network is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network, and in particular to a method, device and storage medium for generating a fault inspection strategy for a distribution network. Background Art

[0002] When a distribution line fails and trips, in order to ensure the normal operation of the distribution line and the normal power consumption of users' electrical equipment, operation and maintenance personnel are required to conduct fault inspections on the distribution network.

[0003] At present, when conducting fault inspections, maintenance personnel are usually randomly dispatched to the site to check whether the line is damaged and use relevant test equipment to evaluate the electrical performance of the line. However, when maintenance personnel are not familiar with the structure of the distribution network and the geographical environment where the equipment is located, they cannot quickly and effectively plan the inspection route, resulting in low efficiency of fault inspections in the distribution network.

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

[0005] The embodiments of the present invention provide a method, a device and a storage medium for generating a fault inspection strategy of a distribution network, so as to at least solve the technical problem of low efficiency of fault inspection of the distribution network.

[0006] According to one aspect of the embodiment of the invention, a method for generating a fault inspection strategy of a distribution network is provided. The method comprises: when the distribution network is in a fault, obtaining the single-line diagram data, geographic location data, and initial state data of multiple hardware devices in the distribution network, wherein the single-line diagram data is obtained through the electrical connection diagram of the distribution network; determining the initial topology structure of the distribution network based on the single-line diagram data and the geographic location data, wherein the initial topology structure is used to characterize the connection relationship between multiple hardware devices; inputting the initial topology structure and the initial state data into the target topology model of the distribution network for analysis, obtaining multiple areas to be inspected and devices to be inspected of the distribution network, wherein the target topology model is used to predict the fault point of the distribution network; based on the target inspection resource information of the distribution network, the target inspection constraint condition of the distribution network and the target inspection function of the distribution network, inspecting the areas to be inspected and the devices to be inspected, and obtaining the inspection results, wherein the target inspection resource information is obtained by constraining the initial inspection resource information of the distribution network through the target inspection constraint condition, and the target inspection function is used to characterize the inspection target to be achieved by inspecting the areas to be inspected and the devices to be inspected; generating the fault inspection strategy of the distribution network based on the inspection results.

[0007] Optionally, based on the target patrol resource information of the distribution network, the target patrol constraints of the distribution network and the target patrol function of the distribution network, the patrol area and the equipment to be patrolled are patrolled to obtain patrol results, including: based on the target patrol resource information, the target patrol constraints and the target patrol function, multiple patrol groups and multiple target patrol paths of the distribution network are generated; based on the multiple patrol groups and the multiple target patrol paths, the patrol area and the equipment to be patrolled are patrolled to obtain patrol results.

[0008] Optionally, the initial topology structure and the initial state data are input into the target topology model of the distribution network for analysis to obtain multiple areas to be inspected and devices to be inspected of the distribution network, including: inputting the initial topology structure and the initial state data into the feature representation layer of the target topology model to obtain first structure feature data and first state feature data; inputting the first structure feature data and the first state feature data into the embedding layer of the target topology model to obtain second structure feature data and second state feature data; inputting the second structure feature data and the second state feature data into the attention mechanism of the target topology model to obtain third structure feature data and third state feature data; and determining multiple areas to be inspected and devices to be inspected based on the third structure feature data, the third state feature data and the target topology model.

[0009] Optionally, based on the third structural characteristic data, the third state characteristic data and the target topology model, multiple areas to be inspected and devices to be inspected are determined, including: inputting the third structural characteristic data and the third state characteristic data into the target topology model to obtain multiple initial inspection areas; using the target topology model to obtain multiple fault occurrence probabilities of multiple initial inspection devices in the initial inspection areas; based on the multiple fault occurrence probabilities, multiple areas to be inspected and devices to be inspected are determined.

[0010] Optionally, based on multiple fault occurrence probabilities, multiple areas to be inspected and devices to be inspected are determined, including: based on multiple fault occurrence probabilities, multiple target inspection devices are selected from multiple initial inspection devices; based on the multiple target inspection devices, multiple devices to be inspected and multiple areas to be inspected corresponding to the multiple devices to be inspected are determined.

[0011] Optionally, the method also includes: extracting features from the initial topology structure to obtain topology feature information of the initial topology structure; updating the initial topology model of the distribution network using the topology feature information to obtain a target topology model, wherein the initial topology model is obtained through the historical topology structure of the distribution network and the historical status data of each hardware device.

[0012] Optionally, the initial topology structure includes: device information of the hardware device and connection information of the hardware device.

[0013] According to one aspect of an embodiment of the present invention, a device for generating a fault inspection strategy for a distribution network is provided, the device also comprising: a first acquisition unit, for acquiring single-line diagram data, geographic location data, and initial state data of a plurality of hardware devices in the distribution network when the distribution network is in a fault state, wherein the single-line diagram data is obtained through an electrical connection diagram of the distribution network; a determination unit, for determining an initial topological structure of the distribution network based on the single-line diagram data and the geographic location data, wherein the initial topological structure is used to characterize the connection relationship between the plurality of hardware devices; an input unit, for inputting the initial topological structure and the initial state data into a target topological model of the distribution network In order to obtain multiple inspection areas and equipment to be inspected of distribution networks, the target topology model is used to predict the fault points of the distribution network; the second acquisition unit is used to inspect the inspection areas and equipment to be inspected based on the target inspection resource information of the distribution network, the target inspection constraint conditions of the distribution network and the target inspection function of the distribution network to obtain the inspection results, wherein the target inspection resource information is obtained by constraining the initial inspection resource information of the distribution network through the target inspection constraint conditions, and the target inspection function is used to characterize the inspection goals to be achieved when inspecting the inspection areas and equipment to be inspected; the generation unit is used to generate a fault inspection strategy for the distribution network based on the inspection results.

[0014] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided. 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 in the embodiment of the present invention.

[0015] 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 the method in the embodiment of the present invention is executed when the program is run.

[0016] According to another aspect of the embodiments of the present invention, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the method in the embodiments of the present invention is implemented.

[0017] According to another aspect of an embodiment of the present invention, an electronic device is provided, the electronic device comprising a processor and a memory for storing instructions executable by the processor, wherein the processor is configured to execute the instructions to implement the method in the embodiment of the present invention.

[0018] In an embodiment of the present invention, when the distribution network is in a faulty state, single-line diagram data, geographic location data, and initial state data of multiple hardware devices in the distribution network are obtained, wherein the single-line diagram data is obtained through the electrical connection diagram of the distribution network; based on the single-line diagram data and the geographic location data, the initial topology structure of the distribution network is determined, wherein the initial topology structure is used to characterize the connection relationship between the multiple hardware devices; the initial topology structure and the initial state data are input into a target topology model of the distribution network for analysis, and multiple areas to be inspected and devices to be inspected of the distribution network are obtained, wherein the target topology model is used to predict the fault point of the distribution network; based on the target inspection resource information of the distribution network, the target inspection constraint condition of the distribution network and the target inspection function of the distribution network, the areas to be inspected and the devices to be inspected are inspected to obtain inspection results, wherein the target inspection resource information is obtained by constraining the initial inspection resource information of the distribution network through the target inspection constraint condition, and the target inspection function is used to characterize the inspection target to be achieved when inspecting the areas to be inspected and the devices to be inspected; based on the inspection results, a fault inspection strategy for the distribution network is generated. That is to say, the embodiment of the present invention can first obtain the single-line diagram data, geographic location data, and initial state data of multiple hardware devices in the distribution network when the distribution network is in a fault state, and then determine the initial topology structure of the distribution network based on the single-line diagram data and the geographic location data. Then, the initial topology result and initial state data obtained above are input into the target topology model of the distribution network for analysis, so as to obtain multiple areas to be inspected and devices to be inspected of the distribution network. According to the target inspection resource information of the distribution network, the target inspection constraint conditions of the distribution network and the target inspection function of the distribution network, the areas to be inspected and devices to be inspected obtained above can be inspected to obtain inspection results. Finally, according to the inspection results, the purpose of generating a fault inspection strategy for the distribution network is achieved. Taking into account that after the initial topological structure of the distribution network is determined based on the obtained single-line diagram data and geographic location data, multiple areas to be inspected and devices to be inspected can be obtained based on the initial topological structure and initial state data and by using the target topological model, so that according to the target inspection resource information, target inspection constraints and target inspection functions, the areas to be inspected and the devices to be inspected can be inspected, thereby achieving the purpose of determining the fault inspection strategy of the distribution network, solving the technical problem of low efficiency of fault inspection of the distribution network, and achieving the technical effect of improving the efficiency of fault inspection of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. 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:

[0020] Figure 1is a flow chart of a method for generating a fault inspection strategy for a distribution network according to an embodiment of the present invention;

[0021] Figure 2 is a flowchart of a fault inspection and dispatching method based on a graph neural network according to an embodiment of the present invention;

[0022] Figure 3 is a schematic diagram of a fault inspection and dispatching system based on a graph neural network according to an embodiment of the present invention;

[0023] Figure 4 It is a schematic diagram of a device for generating a fault inspection strategy for a distribution network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] 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.

[0025] 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, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] According to an embodiment of the present invention, a method for generating a fault patrol strategy for a distribution network is provided. It should be noted that in the flowchart of the accompanying drawings, the steps shown therein 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 may be executed in an order different from that shown here.

[0027] The following is an introduction to a method for generating a fault inspection strategy for a distribution network according to an embodiment of the present invention.

[0028] Figure 1is a flow chart of a method for generating a fault inspection strategy for a distribution network according to an embodiment of the present invention. Figure 1 As shown, the method for generating the fault inspection strategy of the distribution network may include the following steps:

[0029] Step S101, when the distribution network is in a fault state, obtaining single-line diagram data, geographic location data, and initial status data of multiple hardware devices in the distribution network.

[0030] In the technical solution provided in the above step S101 of the present invention, when the distribution network is in a fault, the single-line diagram data, geographic location data and initial state data of multiple hardware devices in the distribution network can be obtained. Among them, the initial state data of multiple hardware devices can be the device state data of each device after the distribution network trips due to a fault, such as the opening and closing state of the switch, the current and voltage of the line, etc. The hardware device can be referred to as a device for short.

[0031] Optionally, the single-line diagram data is obtained through an electrical connection diagram of the distribution network. The hardware device may be an electrical device in the distribution network. The geographic location data may be obtained through geographic location information, and the geographic location information may be obtained through a geographic information system (GIS) diagram. The geographic location data may be referred to as GIS data.

[0032] For example, the electrical connection diagram of the distribution network is collected, which usually includes multiple devices (such as transformers, switches, cables, etc.) and the connection relationship between multiple devices, so as to obtain the single-line diagram data of the distribution network; at the same time, the geographical location information of the distribution network-related equipment is obtained, including longitude and latitude, altitude, etc., as well as the spatial relationship between the equipment, so as to obtain the GIS data of the distribution network.

[0033] Step S102: determining an initial topology structure of the distribution network based on the single-line diagram data and the geographic location data.

[0034] In the technical solution provided in the above step S102 of the present invention, the initial topology structure of the distribution network can be determined based on the obtained single-line diagram data and geographic location data, wherein the initial topology structure is used to characterize the connection relationship between multiple hardware devices, and can be the distribution network topology structure data of the post-tripping section when a fault occurs in the distribution network and the network trips.

[0035] Optionally, the obtained single-line diagram data and geographic location data are integrated to obtain integrated data; the integrated data is preprocessed to obtain preprocessed integrated data; based on the preprocessed integrated data, an initial topological structure is determined, wherein the initial topological structure can be called a graph data structure, in which nodes can be used to represent devices, and edges can be used to represent connections between devices.

[0036] For example, the single-line diagram data is integrated with the GIS data to ensure that each device in the diagram has both electrical connection information and geographic location information; the integrated data is preprocessed, including cleaning, denoising, and standardization, to ensure data quality, thereby obtaining preprocessed integrated data; based on the preprocessed integrated data, a unified graph data structure is created, and the nodes in the graph data structure represent the devices, and the edges represent the connections between the devices (electrical connections or spatial proximity relationships).

[0037] Step S103, inputting the initial topology structure and the initial state data into the target topology model of the distribution network for analysis, and obtaining a plurality of areas to be inspected and devices to be inspected of the distribution network.

[0038] In the technical solution provided in the above step S103 of the present invention, after obtaining the initial topology structure and initial state data, the initial topology structure and initial state data can be input into the target topology model of the distribution network for analysis, so as to achieve the purpose of obtaining the areas to be inspected and the equipment to be inspected of multiple distribution networks.

[0039] Optionally, the target topology model is used to predict the fault point of the distribution network. The target topology model can be called a target graph neural network (GNN) model. The area to be inspected can be an area affected by the fault. The device to be inspected can be a device affected by the fault.

[0040] Alternatively, a graph neural network is a deep learning framework that uses neural networks to process graph-structured data, aiming to extract and discover features and patterns from complex graph-structured data to meet the needs of various graph learning tasks such as clustering, classification, prediction, segmentation, generation, etc.

[0041] For example, when a fault trip occurs, the distribution network topology data of the post-trip section and the equipment status data of each device in the post-trip section are obtained, and the distribution network topology data and equipment status data of the post-trip section are input into the target graph neural network model to obtain the areas and equipment affected by the fault.

[0042] It should be noted that this is only a preferred implementation method for obtaining multiple areas to be inspected and devices to be inspected of distribution networks, and the process and method for obtaining multiple areas to be inspected and devices to be inspected are not specifically limited. As long as the initial topology structure and initial state data are input into the target topology model for analysis to obtain multiple areas to be inspected and devices to be inspected, the process and method are within the scope of protection of the present invention and will not be listed here.

[0043] Step S104, based on the target patrol resource information of the distribution network, the target patrol constraint conditions of the distribution network and the target patrol function of the distribution network, patrol the area to be patrolled and the equipment to be patrolled to obtain patrol results.

[0044] In the technical solution provided in the above step S104 of the present invention, according to the target patrol resource information of the distribution network, the target patrol constraint conditions of the distribution network and the target patrol function of the distribution network, it is possible to patrol the patrol area and the patrol equipment to be patrolled, so as to obtain the patrol result. Among them, patrol refers to checking the operating status and environmental changes of the distribution line, timely discovering problems that may threaten the safe and stable operation of the power grid, and avoiding failures.

[0045] Optionally, the inspection results are used to characterize the operating status of the distribution network and the health status of the equipment, and the inspection results can help operation and maintenance personnel to promptly discover problems in the distribution network and take measures to resolve them, thereby ensuring the normal operation of the distribution network and the safe and stable operation of the equipment.

[0046] Optionally, the target patrol resource information is obtained by constraining the initial patrol resource information of the distribution network through target patrol constraint conditions, and the target patrol function is used to characterize the patrol target to be achieved when patrolling the area to be patrolled and the equipment to be patrolled.

[0047] Optionally, the initial patrol resource information may be the decision variables initially set, specifically referring to the areas and equipment that need to be patrolled after the tripping, and the conditions of different human and material resources, such as available human resources (such as the number and skill level of patrol personnel), material resources (such as the number, type, and load capacity of vehicles, the number and endurance of drones, etc.), terrain characteristics of the patrol area, road conditions, traffic restrictions, etc. Based on the decision variables obtained above, a reasonable path can be planned.

[0048] Optionally, the target inspection constraint condition may be a pre-set inspection constraint condition, which may be referred to as a constraint condition for short, for example, each group requires at least two operation and maintenance personnel, each group requires a corresponding vehicle, each group must inspect the overhead lines (above the sky) and the cables (underground), and the overhead lines are inspected by drones, and the cables are inspected by people, each group needs to be equipped with at least one operation and maintenance personnel who is proficient in controlling the flight of drones, time restrictions are imposed (such as the inspection task must be completed within a certain time), safety restrictions are imposed (such as avoiding dangerous areas), etc. It should be noted that the vehicle corresponding to each group is only a means of transportation, which is responsible for picking up and dropping off the operation and maintenance personnel to the areas that need to be inspected.

[0049] Optionally, the target patrol function may be a pre-set target patrol function, which may be referred to as the target function, for example, minimizing the total patrol time, maximizing the patrol efficiency, or balancing the workload of each group (i.e., each group completes the patrol work in the same time), etc. It should be noted that only examples are given here to illustrate the target patrol function, and the target patrol function is not specifically limited.

[0050] Step S105: generating a fault inspection strategy for the distribution network based on the inspection results.

[0051] In the technical solution provided in the above step S105 of the present invention, after obtaining the inspection result, a fault inspection strategy for the distribution network can be generated, wherein the inspection result is used to characterize whether the area to be inspected and the equipment to be inspected are fault points.

[0052] Optionally, in response to the inspection result that the area to be inspected and the equipment to be inspected are fault points, a fault inspection strategy needs to be generated so that the fault point can be accurately inspected, thereby improving the fault inspection efficiency of the distribution network.

[0053] In the above steps S101 to S105 of the present invention, when the distribution network is in a fault, the single-line diagram data, geographic location data, and initial state data of multiple hardware devices in the distribution network can be first obtained, and then the initial topology structure of the distribution network can be determined based on the single-line diagram data and the geographic location data. The initial topology result and initial state data obtained above are then input into the target topology model of the distribution network for analysis, so as to obtain multiple areas to be inspected and devices to be inspected of the distribution network. According to the target inspection resource information of the distribution network, the target inspection constraint conditions of the distribution network and the target inspection function of the distribution network, the areas to be inspected and devices to be inspected obtained above can be inspected to obtain inspection results. Finally, according to the inspection results, the purpose of generating a fault inspection strategy for the distribution network is achieved. Taking into account that after the initial topological structure of the distribution network is determined based on the obtained single-line diagram data and geographic location data, multiple areas to be inspected and devices to be inspected can be obtained based on the initial topological structure and initial state data and by using the target topological model, so that according to the target inspection resource information, target inspection constraints and target inspection functions, the areas to be inspected and the devices to be inspected can be inspected, thereby achieving the purpose of determining the fault inspection strategy of the distribution network, solving the technical problem of low efficiency of fault inspection of the distribution network, and achieving the technical effect of improving the efficiency of fault inspection of the distribution network.

[0054] The above method of this embodiment is further introduced below.

[0055] As an optional implementation method, based on the target patrol resource information of the distribution network, the target patrol constraints of the distribution network and the target patrol function of the distribution network, the patrol area and the equipment to be patrolled are patrolled to obtain patrol results, including: based on the target patrol resource information, the target patrol constraints and the target patrol function, multiple patrol groups and multiple target patrol paths of the distribution network are generated; based on the multiple patrol groups and the multiple target patrol paths, the patrol area and the equipment to be patrolled are patrolled to obtain patrol results.

[0056] In this embodiment, a heuristic optimization algorithm is used to generate multiple inspection groups of the distribution network and multiple target inspection paths corresponding to the inspection groups according to the target inspection resource information, target inspection constraints and target inspection functions. Then, based on the multiple inspection groups and multiple target inspection paths obtained above, the inspection area and the equipment to be inspected can be inspected to obtain the inspection results.

[0057] Optionally, a target algorithm is selected from multiple algorithms as a heuristic optimization algorithm, such as a genetic algorithm, a particle swarm algorithm, an ant colony algorithm, etc. Since the patrol division of labor and path planning problem involves complex combinatorial optimization and a large amount of calculation, it is suitable to use a heuristic optimization algorithm, which can find an approximate optimal solution to the problem within a reasonable time.

[0058] Optionally, the patrol group may be referred to as a group. The target patrol path may be referred to as a patrol route. The target patrol path of the patrol group may be an optimal patrol route corresponding to the patrol group.

[0059] Optionally, a heuristic optimization algorithm can be used to perform patrol division and path planning based on decision variables, constraints and objective functions, and multiple groups and the optimal patrol route corresponding to each group can be obtained, thereby achieving the purpose of obtaining patrol results.

[0060] In this embodiment, after the possible fault points are determined, combined with the current human and material resources, it is determined that several groups should be divided to jointly patrol these possible fault points, wherein each group has a corresponding optimal patrol route. For example, it is divided into three groups, group A patrols the 1st to 3rd fault points, group B patrols the 4th to 7th fault points, and group C patrols the 8th to 10th fault points. The above process is as follows: a certain number of initial solutions (i.e., patrol division of labor and path planning schemes) are randomly generated; each initial solution is evaluated according to the objective function and constraints, and its fitness value is calculated.

[0061] Furthermore, after obtaining the fitness value of each initial solution, iterative optimization can be performed. The process of iterative optimization: selection operation: select some excellent solutions as parents according to the fitness value; crossover (or mutation) operation: crossover (or mutation) the parent solution to generate new child solutions; fitness re-evaluation: fitness evaluation of the newly generated child solutions; update solution set: compare the newly generated child solutions with the solutions in the current solution set, and retain the solutions with higher fitness.

[0062] In addition, after updating the solution set, it is necessary to determine whether the number of iterations parameter at this time meets the termination condition. When the number of iterations parameter reaches the preset number of iterations or the quality of the current solution is no longer significantly improved, the algorithm terminates; check the optimal solution found by the heuristic optimization algorithm, that is, the optimal patrol division of labor and path planning scheme, and verify the optimal solution according to the actual situation, and make fine adjustments if necessary; according to the final patrol division of labor and path planning scheme, assign corresponding operation and maintenance personnel to perform corresponding patrol work.

[0063] As an optional implementation method, the initial topology structure and initial state data are input into the target topology model of the distribution network for analysis to obtain multiple areas to be inspected and devices to be inspected of the distribution network, including: inputting the initial topology structure and initial state data into the feature representation layer of the target topology model to obtain first structure feature data and first state feature data; inputting the first structure feature data and the first state feature data into the embedding layer of the target topology model to obtain second structure feature data and second state feature data; inputting the second structure feature data and the second state feature data into the attention mechanism of the target topology model to obtain third structure feature data and third state feature data; and determining multiple areas to be inspected and devices to be inspected based on the third structure feature data, the third state feature data and the target topology model.

[0064] In this embodiment, after obtaining the initial topological structure and the initial state data, the initial topological structure and the initial state data can be input into the feature representation layer of the target topological model to obtain the first structural feature data and the first state feature data, and then the first structural feature data and the first state feature data are input into the embedding layer of the target topological model. The first structural feature data and the first state feature data can be processed in a low-dimensional manner to obtain the second structural feature data and the second state feature data, and then the second structural feature data and the second state feature data are input into the attention mechanism of the target topological model to obtain the third structural feature data and the third state feature data, and finally, according to the third structural feature data, the third state feature data and the target topological model, a plurality of areas to be inspected and devices to be inspected can be determined.

[0065] Optionally, the feature representation layer can be a feature extraction layer, which is used to perform feature extraction processing on the initial topological structure and initial state data, so as to obtain corresponding first structural feature data and first state feature data. Among them, the first structural feature data includes: node features (such as device type, capacity, geographical location, etc.), and edge features (such as connection type, distance, etc.). The first state feature data can be the state data of the hardware device, such as the opening and closing state of the switch. It should be noted that only the first structural feature data and the first state feature data are illustrated here.

[0066] For example, the distribution network topology data and equipment status data after the tripping section are input into the target graph neural network model, and the target graph neural network model is used to perform feature extraction on the distribution network topology data and equipment status data after the tripping section to obtain corresponding feature data, such as node features (including equipment type, capacity, geographical location, etc.) and edge features (including connection type, distance, etc.); then, the embedding layer in the target graph neural network model is used to convert these features into low-dimensional vector representations for subsequent processing.

[0067] Furthermore, under the action of the graph convolution layer or attention mechanism in the target graph neural network model, the features of nodes and edges will be propagated and aggregated in the network. Specifically, each node aggregates the information of its neighbor nodes (including direct neighbors and indirect neighbors reached through multi-layer propagation) to update the representation of its own node. This process is repeated many times to capture the complex relationships and deep features in the distribution network.

[0068] For another example, the target graph neural network model will analyze the impact range of fault tripping on the distribution network to obtain the fault impact analysis results. For example, by comparing the changes in node representation and edge representation before and after the fault, multiple areas and devices affected by the fault can be identified.

[0069] As an optional implementation method, based on the third structural characteristic data, the third state characteristic data and the target topology model, multiple areas to be inspected and devices to be inspected are determined, including: inputting the third structural characteristic data and the third state characteristic data into the target topology model to obtain multiple initial inspection areas; using the target topology model to obtain multiple fault occurrence probabilities of multiple initial inspection devices in the initial inspection areas; based on the multiple fault occurrence probabilities, multiple areas to be inspected and devices to be inspected are determined.

[0070] In this embodiment, the obtained third structural characteristic data and third state characteristic data are input into the target topology model to obtain multiple initial patrol areas, and then the target topology model is used to obtain multiple fault occurrence probabilities of multiple initial patrol devices in the initial patrol area, and according to the multiple fault occurrence probabilities obtained above, multiple areas to be patrolled and devices to be patrolled can be determined. Among them, the multiple areas to be patrolled and devices to be patrolled are the areas and devices that need to be patrolled that are finally determined.

[0071] For example, based on the fault impact analysis results obtained above, the target graph neural network model will determine which areas are affected by the fault tripping. These areas may include a collection of equipment such as lines, switches, transformers, etc. near the fault point. Then, using the target graph neural network model, the failure probability of multiple hardware devices can be determined. Based on the failure probability, the areas and equipment that need to be inspected can be finally determined.

[0072] As an optional implementation method, based on multiple fault probabilities, multiple areas to be inspected and devices to be inspected are determined, including: based on multiple fault probabilities, multiple target inspection devices are selected from multiple initial inspection devices; based on the multiple target inspection devices, multiple devices to be inspected and multiple areas to be inspected corresponding to the multiple devices to be inspected are determined.

[0073] In this embodiment, after obtaining multiple fault occurrence probabilities, multiple target patrol devices can be selected from multiple initial patrol devices, and then multiple devices to be patrolled and patrol areas corresponding to the devices to be patrolled can be determined based on the multiple target patrol devices. The target patrol devices can be called key patrol devices.

[0074] Optionally, multiple fault occurrence probabilities are sorted from large to small to obtain multiple sorted fault occurrence probabilities; from the sorted multiple fault occurrence probabilities, target inspection devices corresponding to the first target number of fault occurrence probabilities are selected, wherein the target number can be a preset number, for example, when the target number is 3, then devices corresponding to the first 3 fault occurrence probabilities can be selected from the sorted multiple fault occurrence probabilities as key inspection devices.

[0075] Optionally, after the target graph neural network model determines which areas are affected by the fault tripping, the target graph neural network model will further screen out the equipment in the area that needs to be inspected. These equipment may be the direct cause of the fault or a potential risk point, and the operation and maintenance personnel need to inspect and handle them first. For example, identify the probability of failure of each device in the area, and then select the top few devices in the probability of failure as the equipment that needs to be inspected. After this process, the areas and equipment that need to be inspected after the tripping are finally determined.

[0076] For another example, suppose there are multiple target graph neural network models, each of which will output the areas and equipment that need to be inspected after the tripping. By combining the areas and equipment that need to be inspected after the tripping output by each target graph neural network model, the areas and equipment that need to be inspected in the end can be obtained. For example, from the areas and equipment that need to be inspected after the tripping output by each target graph neural network model, the areas and equipment with higher prediction probabilities are selected as the areas and equipment that need to be inspected in the end.

[0077] As an optional implementation mode, the method also includes: extracting features from the initial topology structure to obtain topology structure feature information of the initial topology structure; using the topology structure feature information to update the initial topology model of the distribution network to obtain a target topology model, wherein the initial topology model is obtained through the historical topology structure of the distribution network and the historical status data of each hardware device.

[0078] In this embodiment, the initial topology structure may be feature extracted to obtain topology structure feature information of the initial topology structure, and then the initial topology model of the distribution network may be updated using the topology structure feature information obtained above to obtain the target topology model.

[0079] Optionally, feature extraction is performed on the node information and edge information in the initial topological structure. Features are extracted for each node (device), such as device type, device capacity, device status (in operation or under maintenance), geographic location (latitude and longitude), etc.; features are extracted for each edge (connection), such as connection type (electrical connection or spatial connection), connection distance (spatial connection), current / voltage level (electrical connection), etc.

[0080] Optionally, the topological structure feature information includes node feature information and edge feature information, the node feature information can be referred to as node feature, and the edge feature information can be referred to as edge feature. The initial topological model can be an untrained graph neural network model, and the target topological model can be a trained graph neural network model, such as Graph Convolutional Networks (GCN), Graph Attention Networks (GAT), or Graph Isomorphism Network (GIN). It should be noted that only the graph neural network model is exemplified here.

[0081] Optionally, after obtaining the node (device) features and edge (connection) features, the selected appropriate graph neural network model is iteratively trained to obtain a trained graph neural network model. For example, the node (device) features and edge (connection) features are input into the selected graph neural network model, and the node (device) features and edge (connection) features are learned and embedded through the multi-layer graph convolution layer or attention mechanism in the selected graph neural network model to capture the complex topological structure of the distribution network and the deep relationship between devices. After multiple rounds of training, a trained graph neural network model can be obtained. Among them, each node and edge in the network updates its representation through the information of its neighbors (connected nodes or edges).

[0082] Furthermore, the trained graph neural network model is parameter-tuned to obtain a target graph neural network model, which can be used for fault prediction. It should be noted that in order to ensure the accuracy of fault prediction, different types of target graph neural network models can be trained based on different types of graph neural network models.

[0083] As an optional implementation manner, the initial topology structure includes: device information of the hardware device and connection information of the hardware device.

[0084] In this embodiment, the initial topology structure may include: device information of the hardware device and connection information of the hardware device, wherein the device information of the hardware device may be node information and the connection information of the hardware device may be edge information.

[0085] Optionally, the device information may include: device type, device capacity, device status (in operation or under maintenance), and geographic location (latitude and longitude), etc. The connection information may include: connection type (electrical connection or spatial connection), connection distance (spatial connection), and current level or voltage level (electrical connection), etc. It should be noted that only examples are given here to illustrate the device information and connection information, and the device information and connection information are not specifically limited.

[0086] In this embodiment, when the distribution network is in a faulty state, the single-line diagram data, geographic location data, and initial state data of multiple hardware devices in the distribution network can be first obtained, and then the initial topology structure of the distribution network can be determined based on the single-line diagram data and the geographic location data. The initial topology results and initial state data obtained above are then input into the target topology model of the distribution network for analysis, so as to obtain multiple areas to be inspected and devices to be inspected of the distribution network. According to the target inspection resource information of the distribution network, the target inspection constraints of the distribution network, and the target inspection function of the distribution network, the areas to be inspected and devices to be inspected obtained above can be inspected to obtain inspection results. Finally, according to the inspection results, the purpose of generating a fault inspection strategy for the distribution network is achieved. Taking into account that after the initial topological structure of the distribution network is determined based on the obtained single-line diagram data and geographic location data, multiple areas to be inspected and devices to be inspected can be obtained based on the initial topological structure and initial state data and by using the target topological model, so that according to the target inspection resource information, target inspection constraints and target inspection functions, the areas to be inspected and the devices to be inspected can be inspected, thereby achieving the purpose of determining the fault inspection strategy of the distribution network, solving the technical problem of low efficiency of fault inspection of the distribution network, and achieving the technical effect of improving the efficiency of fault inspection of the distribution network.

[0087] The technical solution of the embodiment of the present invention is illustrated below in conjunction with preferred implementation modes.

[0088] There are many reasons for the fault tripping of distribution lines, mainly including overload operation causing the current to exceed the rated carrying capacity of the line or equipment, such as high current impact caused by short circuit faults, leakage caused by insulation aging or damage, natural environmental factors (lightning hitting the line) causing line disconnection and contact, and human factors (such as construction misoperation or animal activities) damaging the line, etc. These factors may cause the protection device to detect abnormal conditions and automatically cut off the circuit, that is, trip, to prevent more serious consequences such as equipment damage or fire.

[0089] Therefore, when a distribution line fails and trips, in order to ensure the normal operation of the distribution line and the normal power consumption of the user's electrical equipment, operation and maintenance personnel are required to conduct fault inspections on the distribution network in order to quickly find and isolate the fault point and restore power supply.

[0090] At present, when conducting fault inspections, maintenance personnel are usually randomly dispatched to the site to check whether the line is damaged, and use relevant test equipment to evaluate the electrical performance of the line, including insulation and grounding status, while paying close attention to whether there are potential safety hazards in the surrounding environment. However, when maintenance personnel are not familiar with the structure of the distribution network and the geographical environment where the equipment is located, they cannot quickly and effectively plan the inspection route, resulting in the technical problem of low efficiency of fault inspections in the distribution network.

[0091] Therefore, in order to solve the above problems, the present invention proposes a fault inspection and dispatching method based on graph neural network. The method can be based on the single-line diagram of the line and the equipment location information. After a fault occurs, it can first determine which equipment is a possible fault point that needs to be inspected, and then establish appropriate groups based on the current personnel and material conditions. Finally, the optimal inspection route is planned for each group (ideally, each group completes the inspection work in the same time) to reduce the fault investigation time, thereby solving the technical problem of low fault inspection efficiency of the distribution network and achieving the technical effect of improving the fault inspection efficiency of the distribution network.

[0092] In this embodiment, Figure 2 The present invention is a flowchart of a method for fault inspection and dispatching based on a graph neural network according to an embodiment of the present invention. The method for fault inspection and dispatching based on a graph neural network mainly includes the following steps:

[0093] Step S201, based on the single-line diagram and geographic information system diagram of the distribution network, the topological structure of the distribution network is learned by using graph neural network modeling.

[0094] In this embodiment, the single-line diagram and geographic information system diagram of the distribution network can be collected, so as to determine the topological structure of the distribution network by using a graph neural network based on the single-line diagram and geographic information system diagram obtained above.

[0095] Optionally, collect the electrical connection diagram of the distribution network, which usually includes multiple devices (such as transformers, switches, cables, etc.) and the connection relationship between multiple devices, so as to obtain the single-line diagram data of the distribution network; at the same time, obtain the geographical location information of the distribution network related equipment, including longitude and latitude, altitude, etc., as well as the spatial relationship between the equipment, so as to obtain the GIS data of the distribution network.

[0096] Optionally, the single-line diagram data is integrated with the GIS data to ensure that each device in the diagram has both electrical connection information and geographic location information; the integrated data is preprocessed, including cleaning, denoising, standardization, etc., to ensure data quality, thereby obtaining preprocessed integrated data; based on the preprocessed integrated data, a unified graph data structure is created, and the nodes in the graph data structure represent devices, and the edges represent connections between devices (electrical connections or spatial proximity relationships).

[0097] Optionally, features are extracted from each node (device) to obtain node features, such as equipment type, equipment capacity, equipment status (in operation or under maintenance), geographic location (latitude and longitude), etc.; features are extracted from each edge (connection) to obtain edge features, such as connection type (electrical connection or spatial connection), connection distance (spatial connection), current / voltage level (electrical connection), etc.

[0098] Optionally, select a suitable graph neural network model, such as GCN, GAT, or GIN. It should be noted that these graph neural network models usually include multiple graph convolutional layers for learning feature representations of nodes and edges.

[0099] Optionally, after obtaining the node (device) features and edge (connection) features, the selected appropriate graph neural network model is iteratively trained to obtain a trained graph neural network model. For example, the node (device) features and edge (connection) features are input into the selected graph neural network model, and the node (device) features and edge (connection) features are learned and embedded through the multi-layer graph convolution layer or attention mechanism in the selected graph neural network model to capture the complex topological structure of the distribution network and the deep relationship between devices. After multiple rounds of training, a trained graph neural network model can be obtained. Among them, each node and edge in the network updates its representation through the information of its neighbors (connected nodes or edges).

[0100] Optionally, the trained graph neural network model is parameter tuned to obtain a target graph neural network model, which can be used for fault prediction. It should be noted that in order to ensure the accuracy of fault prediction, different types of target graph neural network models can be trained based on different types of graph neural network models.

[0101] Step S202: when a fault trip occurs, the topology of the distribution network is analyzed to determine the areas and equipment that need to be inspected after the trip.

[0102] In this embodiment, after obtaining the topological structure, the topological structure of the distribution network may be analyzed so as to obtain the areas and equipment that need to be inspected in the post-tripping stage.

[0103] Optionally, when a fault trip occurs, the topological structure data of the distribution network after the trip is obtained (wherein nodes represent equipment such as transformers, switches, lines, and edges represent electrical connections or spatial proximity between these devices), as well as the equipment status data of each equipment after the trip, such as the opening and closing status of the switch, the current and voltage conditions of the line, etc.

[0104] Optionally, the distribution network topology data and equipment status data of the post-tripping section are input into the target graph neural network model, and feature extraction is performed on the distribution network topology data and equipment status data of the post-tripping section through the target graph neural network model to obtain corresponding feature data, such as node features (including equipment type, capacity, geographical location, etc.) and edge features (including connection type, distance, etc.); then, the embedding layer in the target graph neural network model is used to convert these features into low-dimensional vector representations for subsequent processing.

[0105] Furthermore, under the action of the graph convolution layer or attention mechanism in the target graph neural network model, the features of nodes and edges will be propagated and aggregated in the network. Specifically, each node aggregates the information of its neighbor nodes (including direct neighbors and indirect neighbors reached through multi-layer propagation) to update the representation of its own node. This process is repeated many times to capture the complex relationships and deep features in the distribution network.

[0106] Optionally, the target graph neural network model analyzes the impact range of the fault tripping on the distribution network to obtain the fault impact analysis results. For example, by comparing the changes in node representation and edge representation before and after the fault, multiple areas and devices affected by the fault can be identified.

[0107] Optionally, based on the above-obtained fault impact analysis results, the target graph neural network model will determine which areas are affected by the fault tripping. These areas may include a collection of equipment such as lines, switches, transformers, etc. near the fault point; after determining the areas affected by the fault, the target graph neural network model will further screen out the equipment that needs to be inspected in the area. These equipment may be the direct cause of the fault or a potential risk point, and the operation and maintenance personnel need to give priority to inspection and processing. For example, identify the probability of failure of each device in the area, and then select the top few devices in the probability of failure as the equipment that needs to be inspected. After this process, the areas and equipment that need to be inspected after the tripping are finally determined.

[0108] For example, assuming that there are multiple target graph neural network models, each target graph neural network model will output the areas and equipment that need to be inspected after the tripping. By combining the areas and equipment that need to be inspected after the tripping output by each target graph neural network model, the areas and equipment that need to be inspected in the end can be obtained. For example, from the areas and equipment that need to be inspected after the tripping output by each target graph neural network model, the areas and equipment with higher prediction probabilities are selected as the areas and equipment that need to be inspected in the end.

[0109] Step S203, based on different human and material resources, use a heuristic optimization algorithm to perform patrol division and route planning on the distribution network to determine the best grouping arrangement and patrol route.

[0110] In this embodiment, according to different human and material resources, and by using a heuristic optimization algorithm, the distribution network can be patrolled and divided and the route planning can be carried out to achieve the purpose of determining the best grouping arrangement and patrol route.

[0111] Optionally, a heuristic optimization algorithm can be used to perform patrol division and path planning based on decision variables, constraints and objective functions to obtain multiple groups and the optimal patrol route corresponding to each group.

[0112] In this embodiment, after the possible fault points are determined, combined with the current human and material resources, it is determined that several groups should be divided to jointly patrol these possible fault points, wherein each group has a corresponding optimal patrol route. For example, it is divided into three groups, group A patrols the 1st to 3rd fault points, group B patrols the 4th to 7th fault points, and group C patrols the 8th to 10th fault points. The above process is as follows: a certain number of initial solutions (i.e., patrol division of labor and path planning schemes) are randomly generated; each initial solution is evaluated according to the objective function and constraints, and its fitness value is calculated.

[0113] Furthermore, after obtaining the fitness value of each initial solution, iterative optimization can be performed. The process of iterative optimization: selection operation: select some excellent solutions as parents according to the fitness value; crossover (or mutation) operation: crossover (or mutation) the parent solution to generate new child solutions; fitness re-evaluation: fitness evaluation of the newly generated child solutions; update solution set: compare the newly generated child solutions with the solutions in the current solution set, and retain the solutions with higher fitness.

[0114] In addition, after updating the solution set, it is necessary to determine whether the number of iterations parameter at this time meets the termination condition. When the number of iterations parameter reaches the preset number of iterations or the quality of the current solution is no longer significantly improved, the algorithm terminates; check the optimal solution found by the heuristic optimization algorithm, that is, the optimal patrol division of labor and path planning scheme, and verify the optimal solution according to the actual situation, and make fine adjustments if necessary; according to the final patrol division of labor and path planning scheme, assign corresponding operation and maintenance personnel to perform corresponding patrol work.

[0115] Optionally, set constraints. For example, each group needs at least two operation and maintenance personnel, each group needs a corresponding vehicle, each group must patrol overhead lines (above the sky) and cables (underground), and the overhead lines are patrolled by drones, and the cables are patrolled by people, each group needs to be equipped with at least one operation and maintenance personnel who is proficient in controlling the flight of drones, time restrictions (such as the patrol task must be completed within a certain time), safety restrictions (such as avoiding dangerous areas), etc. It should be noted that the vehicle corresponding to each group is only a means of transportation, which is responsible for picking up and dropping off operation and maintenance personnel to the areas that need to be patrolled.

[0116] Optionally, decision variables are set, specifically referring to the areas and equipment that need to be inspected after the tripping, as well as the conditions of different human and material resources, such as available human resources (such as the number and skill level of inspectors), material resources (such as the number, type, and load capacity of vehicles, the number and endurance of drones, etc.), terrain characteristics of the inspection area, road conditions, traffic restrictions, etc. Based on the decision variables obtained above, a reasonable path can be planned.

[0117] Optionally, an objective function is set, such as minimizing the total patrol time, maximizing the patrol efficiency, or balancing the workload of each group (ie, each group completes the patrol work within the same time).

[0118] Optionally, select a suitable heuristic optimization algorithm, such as genetic algorithm, particle swarm algorithm, ant colony algorithm, etc. Since the patrol division of labor and path planning problem involves complex combinatorial optimization and large amount of calculation, it is suitable to adopt a heuristic optimization algorithm, which can find an approximate optimal solution to the problem within a reasonable time.

[0119] In an embodiment of the present invention, Figure 3 is a schematic diagram of a fault inspection and dispatching system based on a graph neural network according to an embodiment of the present invention. Figure 3 As shown, the system includes: a host 301 and an extension 302.

[0120] Host 301 is used for data storage and calculation.

[0121] Extension 302 is used to view the group's work assignment plan and optimal patrol route.

[0122] Optionally, the extension may be referred to as a Personal Digital Assistant (PDA) extension.

[0123] In this embodiment, each operation and maintenance personnel can be assigned a corresponding PDA extension. In addition, the algorithm is encapsulated inside the host, and the path planning and task assignment can be completed offline by simply importing the line single-line diagram and GIS diagram of the distribution network into the host Secure Digital (SD) card. The specific internal implementation process is the specific implementation process of the above-mentioned fault inspection and dispatching method based on graph neural network. Moreover, the host can send the corresponding path and task to the PDA extension of the corresponding operation and maintenance personnel, and the operation and maintenance personnel can view the group's dispatching plan and the optimal inspection path at any time through the PDA extension.

[0124] In this embodiment, when the distribution network is in a faulty state, the single-line diagram data, geographic location data, and initial state data of multiple hardware devices in the distribution network can be first obtained, and then the initial topology structure of the distribution network can be determined based on the single-line diagram data and the geographic location data. The initial topology results and initial state data obtained above are then input into the target topology model of the distribution network for analysis, so as to obtain multiple areas to be inspected and devices to be inspected of the distribution network. According to the target inspection resource information of the distribution network, the target inspection constraints of the distribution network, and the target inspection function of the distribution network, the areas to be inspected and devices to be inspected obtained above can be inspected to obtain inspection results. Finally, according to the inspection results, the purpose of generating a fault inspection strategy for the distribution network is achieved. Taking into account that after the initial topological structure of the distribution network is determined based on the obtained single-line diagram data and geographic location data, multiple areas to be inspected and devices to be inspected can be obtained based on the initial topological structure and initial state data and by using the target topological model, so that according to the target inspection resource information, target inspection constraints and target inspection functions, the areas to be inspected and the devices to be inspected can be inspected, thereby achieving the purpose of determining the fault inspection strategy of the distribution network, solving the technical problem of low efficiency of fault inspection of the distribution network, and achieving the technical effect of improving the efficiency of fault inspection of the distribution network.

[0125] According to an embodiment of the present invention, a device for generating a fault patrol strategy for a distribution network is provided. It should be noted that the device for generating a fault patrol strategy for a distribution network can be used to execute a method for generating a fault patrol strategy for a distribution network in Example 1.

[0126] Figure 4 is a schematic diagram of a device for generating a fault inspection strategy for a distribution network according to an embodiment of the present invention, such as Figure 4 As shown, a device 400 for generating a fault inspection strategy for a distribution network may include: a first acquisition unit 401 , a determination unit 402 , an input unit 403 , a second acquisition unit 404 and a generation unit 405 .

[0127] The first acquisition unit 401 is used to acquire single-line diagram data, geographic location data, and initial status data of multiple hardware devices in the distribution network when the distribution network is in a fault state, wherein the single-line diagram data is obtained through the electrical connection diagram of the distribution network.

[0128] The determination unit 402 is used to determine an initial topology structure of the distribution network based on the single-line diagram data and the geographic location data, wherein the initial topology structure is used to characterize the connection relationship between multiple hardware devices.

[0129] The input unit 403 is used to input the initial topology structure and initial state data into the target topology model of the distribution network for analysis to obtain multiple areas and devices to be inspected of the distribution network, wherein the target topology model is used to predict the fault points of the distribution network.

[0130] The second acquisition unit 404 is used to inspect the inspection area and the equipment to be inspected based on the target inspection resource information of the distribution network, the target inspection constraint conditions of the distribution network and the target inspection function of the distribution network to obtain the inspection result, wherein the target inspection resource information is obtained by constraining the initial inspection resource information of the distribution network through the target inspection constraint conditions, and the target inspection function is used to represent the inspection target that needs to be achieved when inspecting the inspection area and the equipment to be inspected.

[0131] The generating unit 405 is used to generate a fault inspection strategy for the distribution network based on the inspection result.

[0132] Optionally, the second acquisition unit 404 includes: a generation module, which is used to generate multiple inspection groups and multiple target inspection paths of the distribution network based on target inspection resource information, target inspection constraints and target inspection functions; a patrol module, which is used to patrol the inspection area and the equipment to be patrolled based on multiple patrol groups and multiple target patrol paths to obtain patrol results.

[0133] Optionally, the input unit 403 includes: a first input module, used to input the initial topological structure and initial state data into the feature representation layer of the target topological model to obtain first structural feature data and first state feature data; a second input module, used to input the first structural feature data and the first state feature data into the embedding layer of the target topological model to obtain second structural feature data and second state feature data; a third input module, used to input the second structural feature data and the second state feature data into the attention mechanism of the target topological model to obtain third structural feature data and third state feature data; a determination module, used to determine multiple areas to be inspected and devices to be inspected based on the third structural feature data, the third state feature data and the target topological model.

[0134] Optionally, the determination module may include: an input submodule, used to input the third structural characteristic data and the third state characteristic data into the target topology model to obtain multiple initial patrol areas; an acquisition submodule, used to use the target topology model to obtain multiple fault occurrence probabilities of multiple initial patrol devices in the initial patrol area; and a determination submodule, used to determine multiple areas to be patrolled and devices to be patrolled based on multiple fault occurrence probabilities.

[0135] Optionally, the determination submodule is also used to select multiple target patrol devices from multiple initial patrol devices based on multiple fault occurrence probabilities; and is also used to determine multiple devices to be patrolled and multiple areas to be patrolled corresponding to the multiple devices to be patrolled based on the multiple target patrol devices.

[0136] Optionally, the device also includes: a third acquisition unit, used to extract features from the initial topology structure to obtain topology feature information of the initial topology structure; an update unit, used to update the initial topology model of the distribution network using the topology feature information to obtain a target topology model, wherein the initial topology model is obtained through the historical topology structure of the distribution network and the historical status data of each hardware device.

[0137] Optionally, the initial topology structure includes: device information of the hardware device and connection information of the hardware device.

[0138] In this embodiment, when the distribution network is in a fault state, the single-line diagram data, geographic location data, and initial state data of multiple hardware devices in the distribution network are obtained by a first acquisition unit, wherein the single-line diagram data is obtained through an electrical connection diagram of the distribution network; the initial topology structure of the distribution network is determined by a determination unit based on the single-line diagram data and the geographic location data, wherein the initial topology structure is used to characterize the connection relationship between multiple hardware devices; the initial topology structure and the initial state data are input into a target topology model of the distribution network for analysis by an input unit, thereby obtaining multiple areas to be inspected and devices to be inspected of the distribution network, wherein the target topology model is used to predict The invention discloses a method for inspecting the fault point of the distribution network; inspecting the inspection area and the equipment to be inspected based on the target inspection resource information of the distribution network, the target inspection constraint conditions of the distribution network and the target inspection function of the distribution network by a second acquisition unit to obtain the inspection result, wherein the target inspection resource information is obtained by constraining the initial inspection resource information of the distribution network through the target inspection constraint conditions, and the target inspection function is used to characterize the inspection target to be achieved when inspecting the inspection area and the equipment to be inspected; generating a fault inspection strategy of the distribution network based on the inspection result by a generation unit to solve the technical problem of low efficiency of fault inspection of the distribution network, thereby achieving the technical effect of improving the efficiency of fault inspection of the distribution network.

[0139] According to an embodiment of the present invention, a computer-readable storage medium is also provided, which includes a stored program, wherein when the program is run by a processor, the device where the readable storage medium is located is controlled to execute the method for generating a fault patrol strategy for a distribution network in the embodiment.

[0140] According to 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 generating a fault inspection strategy for a distribution network in the embodiment is executed.

[0141] According to an embodiment of the present invention, a computer program product is also provided, including a computer program, and when the computer program is executed by a processor, the method for generating a fault inspection strategy for a distribution network in an embodiment of the present invention is implemented.

[0142] According to an embodiment of the present invention, an electronic device is also provided, the electronic device comprising a processor and a memory for storing instructions executable by the processor, wherein the processor is configured to execute instructions to implement the method for generating a fault patrol strategy for a distribution network in an embodiment of the present invention.

[0143] 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.

[0144] 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.

[0145] In the several embodiments provided by the present invention, 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, and there can be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point, 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.

[0146] 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.

[0147] 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.

[0148] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, 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 product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a 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, magnetic disk or optical disk and other media that can store program codes.

[0149] 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 generating a fault inspection strategy for a distribution network, characterized in that: include: When the distribution network is in a fault state, obtaining single-line diagram data, geographic location data, and initial state data of a plurality of hardware devices in the distribution network, wherein the single-line diagram data is obtained through an electrical connection diagram of the distribution network; Determine an initial topology structure of the distribution network based on the single-line diagram data and the geographic location data, wherein the initial topology structure is used to characterize a connection relationship between a plurality of the hardware devices; Inputting the initial topology structure and the initial state data into the target topology model of the distribution network for analysis, and obtaining a plurality of areas to be inspected and devices to be inspected of the distribution network, wherein the target topology model is used to predict the fault points of the distribution network; Based on the target patrol resource information of the distribution network, the target patrol constraint condition of the distribution network and the target patrol function of the distribution network, the area to be patrolled and the equipment to be patrolled are patrolled to obtain patrol results, wherein the target patrol resource information is obtained by constraining the initial patrol resource information of the distribution network through the target patrol constraint condition, and the target patrol function is used to characterize the patrol target to be achieved by patrolling the area to be patrolled and the equipment to be patrolled; Based on the inspection result, a fault inspection strategy for the distribution network is generated.

2. The method according to claim 1, characterized in that Based on the target patrol resource information of the distribution network, the target patrol constraint condition of the distribution network and the target patrol function of the distribution network, patrol the area to be patrolled and the equipment to be patrolled to obtain patrol results, including: Based on the target patrol resource information, the target patrol constraint condition and the target patrol function, generate a plurality of patrol groups and a plurality of target patrol paths of the distribution network; Based on the multiple patrol groups and the multiple target patrol paths, the area to be patrolled and the equipment to be patrolled are patrolled to obtain the patrol result.

3. The method according to claim 1, characterized in that The initial topology structure and the initial state data are input into the target topology model of the distribution network for analysis to obtain a plurality of areas to be inspected and devices to be inspected of the distribution network, including: Inputting the initial topological structure and the initial state data into the feature representation layer of the target topological model to obtain first structure feature data and first state feature data; Inputting the first structural feature data and the first state feature data into the embedding layer of the target topological model to obtain second structural feature data and second state feature data; Inputting the second structural feature data and the second state feature data into the attention mechanism of the target topological model to obtain third structural feature data and third state feature data; Based on the third structural characteristic data, the third state characteristic data and the target topological model, a plurality of the areas to be inspected and the devices to be inspected are determined.

4. The method according to claim 3, characterized in that Based on the third structural feature data, the third state feature data and the target topology model, determining a plurality of areas to be inspected and devices to be inspected includes: Inputting the third structural characteristic data and the third state characteristic data into the target topological model to obtain a plurality of the initial patrol areas; Using the target topology model, obtaining multiple fault occurrence probabilities of multiple initial patrol devices in the initial patrol area; Based on the multiple fault occurrence probabilities, multiple areas to be inspected and multiple devices to be inspected are determined.

5. The method according to claim 4, characterized in that Based on the multiple fault occurrence probabilities, determining multiple areas to be inspected and multiple devices to be inspected includes: Based on the multiple fault occurrence probabilities, selecting multiple target patrol devices from the multiple initial patrol devices; Based on the plurality of target patrol devices, a plurality of devices to be patrolled and a plurality of areas to be patrolled corresponding to the plurality of devices to be patrolled are determined.

6. The method according to claim 1, characterized in that The method further comprises: Extracting features from the initial topological structure to obtain topological structure feature information of the initial topological structure; The initial topology model of the distribution network is updated using the topology structure characteristic information to obtain the target topology model, wherein the initial topology model is obtained through the historical topology structure of the distribution network and the historical status data of each of the hardware devices.

7. The method according to claim 1, characterized in that The initial topology structure includes: device information of the hardware device and connection information of the hardware device.

8. A device for generating a fault inspection strategy for a distribution network, characterized in that: include: A first acquisition unit is used to acquire single-line diagram data, geographic location data, and initial state data of multiple hardware devices in the distribution network when the distribution network is in a fault state, wherein the single-line diagram data is obtained through an electrical connection diagram of the distribution network; A determination unit, configured to determine an initial topology structure of the distribution network based on the single-line diagram data and the geographic location data, wherein the initial topology structure is used to characterize a connection relationship between a plurality of the hardware devices; An input unit, used for inputting the initial topology structure and the initial state data into a target topology model of the distribution network for analysis, to obtain a plurality of areas to be inspected and devices to be inspected of the distribution network, wherein the target topology model is used for predicting fault points of the distribution network; A second acquisition unit is used to patrol the area to be patrolled and the equipment to be patrolled based on the target patrol resource information of the distribution network, the target patrol constraint condition of the distribution network and the target patrol function of the distribution network to obtain a patrol result, wherein the target patrol resource information is obtained by constraining the initial patrol resource information of the distribution network through the target patrol constraint condition, and the target patrol function is used to characterize the patrol target to be achieved by patrolling the area to be patrolled and the equipment to be patrolled; A generating unit is used to generate a fault inspection strategy for the distribution network based on the inspection result.

9. 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 7.

10. An electronic device, characterized in that: The invention comprises one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any one of claims 1 to 7.

Images