Fault Detection Method, Device and Electronic Equipment Applied to Distribution Network

Abstract

Embodiments of the present disclosure disclose a fault detection method, apparatus, and electronic device applied to a distribution network. A specific implementation manner of the method includes: locating at least one node to be detected in the distribution network to obtain a set of information of the nodes to be detected; for the information of the nodes to be detected, perform the following processing steps: collecting real-time node status information; performing fault detection on the nodes to be detected corresponding to the information of the nodes to be detected according to the real-time node status information and the local network topology map included in the information of the nodes to be detected; in response to the node fault probability being greater than a preset node fault probability and the node spread degree being less than a preset node spread degree, initiating a fault warning matching the node fault type; in response to the node fault probability being greater than a preset node fault probability and the node spread degree being greater than or equal to a preset node spread degree, initiating a risk warning for the local distribution network corresponding to the local network topology map included in the information of the nodes to be detected. This implementation manner ensures the stability of the distribution network and reduces the risk of equipment damage.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technologies, and more particularly, to a fault detection method, apparatus, and electronic device applied to a distribution network. Background Art

[0002] With the increasing development of industrial modernization, large-scale distribution networks have emerged accordingly. At the same time, due to the increased complexity of the distribution network, the probability of faults in the distribution network has also increased significantly. How to timely detect faults in the distribution network to ensure the stability of the distribution network, reduce the risk of equipment damage, and reduce the losses caused by power supply interruptions has become an urgent problem to be solved.

[0003] The above information disclosed in this background art section is only used to enhance the understanding of the background of the inventive concept, and thus, it may include information that does not form the prior art known to those of ordinary skill in the art. Summary of the Invention

[0004] The content part of the present disclosure is used to briefly introduce the inventive concepts, which will be described in detail in the following detailed implementation section. The content part of the present disclosure is not intended to identify the key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0005] Some embodiments of the present disclosure propose a fault detection method, apparatus, and electronic device applied to a distribution network to solve one or more of the technical problems mentioned in the above background art section.

[0006] In a first aspect, some embodiments of the present disclosure provide a fault detection method applied to a distribution network. The method includes: locating at least one node to be detected in the distribution network to obtain a set of node information to be detected, where the distribution network is used to distribute the electric energy transmitted by the transmission network to power consumption terminals, and the nodes to be detected and the node information to be detected are in one-to-one correspondence. The node information to be detected in the set of node information to be detected includes: the type of the node to be detected, the location of the node to be detected, and a local network topology diagram, where the local network topology diagram characterizes the topological structure of the local distribution network where the node to be detected is located, and the local distribution network is a subnet of the distribution network; for each piece of node information to be detected in the set of node information to be detected, perform the following processing steps: collect real-time node status information according to the type of the node to be detected and the location of the node to be detected included in the node information to be detected; perform fault detection on the node to be detected corresponding to the node information to be detected according to the real-time node status information and the local network topology diagram included in the node information to be detected, so as to generate node detection information, where the node detection information includes: the type of node fault, the probability of node fault, and the degree of node influence, where the probability of node fault characterizes the occurrence probability of the fault corresponding to the type of node fault, and the degree of node influence characterizes the influence degree of the node to be detected on the equipment in the local distribution network corresponding to the local network topology diagram; in response to the probability of node fault being greater than a preset node fault probability and the degree of node influence being less than a preset degree of node influence, initiate a fault warning matching the type of node fault; in response to the probability of node fault being greater than the preset node fault probability and the degree of node influence being greater than or equal to the preset degree of node influence, initiate a risk warning for the local distribution network corresponding to the local network topology diagram included in the node information to be detected.

[0007] Second aspect, some embodiments of the present disclosure provide a fault detection device applied to a distribution network. The device includes: a positioning unit configured to locate at least one node to be detected in the distribution network to obtain a set of information of the nodes to be detected. The distribution network is used to distribute the electric energy transmitted by the transmission network to the power consumption terminals. Each node to be detected corresponds to the information of the node to be detected. The information of the nodes to be detected in the set of information of the nodes to be detected includes: the type of the node to be detected, the location of the node to be detected, and a local network topology map, where the local network topology map characterizes the topological structure of the local distribution network where the node to be detected is located, and the local distribution network is a subnet of the above-mentioned distribution network; an execution unit configured to perform the following processing steps for each piece of information of the nodes to be detected in the set of information of the nodes to be detected: collect real-time node status information according to the type of the node to be detected and the location of the node to be detected included in the information of the node to be detected; perform fault detection on the node to be detected corresponding to the information of the node to be detected according to the real-time node status information and the local network topology map included in the information of the node to be detected to generate node detection information, where the node detection information includes: the type of node fault, the probability of node fault, and the degree of node impact, where the probability of node fault characterizes the occurrence probability of the fault corresponding to the type of node fault, and the degree of node impact characterizes the influence degree of the node to be detected on the devices in the local distribution network corresponding to the local network topology map; in response to the probability of node fault being greater than a preset probability of node fault and the degree of node impact being less than a preset degree of node impact, initiate a fault warning matching the type of node fault; in response to the probability of node fault being greater than the preset probability of node fault and the degree of node impact being greater than or equal to the preset degree of node impact, initiate a risk warning for the local distribution network corresponding to the local network topology map included in the information of the node to be detected.

[0008] Third aspect, some embodiments of the present disclosure provide an electronic device, including: one or more processors; a storage device on which one or more programs are stored. 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 implementation manner of the first aspect above.

[0009] Fourth aspect, some embodiments of the present disclosure provide a computer-readable medium on which a computer program is stored. When the program is executed by a processor, the method described in any implementation manner of the first aspect above is implemented.

[0010] The above embodiments of the present disclosure have the following beneficial effects: Through the fault detection method applied to the distribution network in some embodiments of the present disclosure, effective fault detection for the distribution network is achieved, ensuring the stability of the distribution network, reducing the risk of equipment damage, and reducing the losses caused by power supply interruption. Specifically, the reasons for faults in the distribution network are as follows: The structure of large-scale distribution networks is complex. When a device in the distribution network fails, it is very likely to affect the overall stability of the distribution network, thereby affecting other devices and causing equipment damage and power supply interruption. Based on this, in some embodiments of the present disclosure, the fault detection method applied to the distribution network first locates at least one node to be detected in the distribution network to obtain a set of information about the nodes to be detected. Among them, the above distribution network is used to distribute the electric energy transmitted by the transmission network to the power consumption terminals, and the nodes to be detected correspond one-to-one with the information about the nodes to be detected. The information about the nodes to be detected in the above set of information about the nodes to be detected includes: the type of the node to be detected, the location of the node to be detected, and the local network topology diagram, where the local network topology diagram characterizes the topological structure of the local distribution network where the node to be detected is located, and the local distribution network is a subnet of the above distribution network. In practice, the network structure of large-scale distribution networks is complex. If the entire distribution network is subjected to fault detection, the detection difficulty is large and the detection cycle is long. Therefore, by locating the key nodes (nodes to be detected) in the distribution network, the detection complexity can be reduced. Secondly, for each piece of information about the nodes to be detected in the above set of information about the nodes to be detected, the following processing steps are performed: First step, according to the type of the node to be detected and the location of the node to be detected included in the information about the nodes to be detected, real-time node status information is collected. By collecting the real-time node status, the current operating status of the device corresponding to the node to be detected can be obtained. Second step, based on the above real-time node status information and the local network topology diagram included in the information about the nodes to be detected, fault detection is performed on the node to be detected corresponding to the information about the nodes to be detected to generate node detection information, where the above node detection information includes: the type of node fault, the probability of node fault, and the degree of node influence. Among them, the probability of node fault characterizes the occurrence probability of the fault corresponding to the type of node fault, and the degree of node influence characterizes the influence degree of the node to be detected on the devices in the local distribution network corresponding to the local network topology diagram. In this way, from the perspective of the node to be detected itself and the influence of the node to be detected on the surrounding devices, the influence of the type of node fault on the surrounding devices is detected. Third step, in response to the above probability of node fault being greater than the preset probability of node fault and the above degree of node influence being less than the preset degree of node influence, a fault warning matching the above type of node fault is initiated. In this way, a single-point warning for the device corresponding to the node to be detected is realized. Fourth step, in response to the above probability of node fault being greater than the above preset probability of node fault and the above degree of node influence being greater than or equal to the above preset degree of node influence, a risk warning for the local distribution network corresponding to the local network topology diagram included in the above information about the nodes to be detected is initiated. In this way, an overall warning for the node to be detected and the local distribution network where it is located is realized.In this way, fault detection can be efficiently achieved, ensuring the stability of the distribution network, reducing the risk of equipment damage, and reducing the losses caused by power supply interruptions. Brief Description of the Drawings

[0011] In combination with the drawings and with reference to the following specific embodiments, the above and other features, advantages, and aspects of the embodiments of the present disclosure will become more apparent. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the elements and elements are not necessarily drawn to scale.

[0012] Figure 1 is a flowchart of some embodiments of a fault detection method applied to a distribution network according to the present disclosure;

[0013] Figure 2 is a schematic structural diagram of some embodiments of a fault detection device applied to a distribution network according to the present disclosure;

[0014] Figure 3 is a schematic structural diagram of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Description of the Embodiments

[0015] The embodiments of the present disclosure will be described in more detail below with reference to the drawings. Although some embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for illustrative purposes and are not used to limit the protection scope of the present disclosure.

[0016] In addition, it should be noted that for the sake of convenience of description, only the parts related to the relevant invention are shown in the drawings. Without conflict, the embodiments in the present disclosure and the features in the embodiments can be combined with each other.

[0017] It should be noted that the concepts such as "first" and "second" mentioned in the present disclosure are only used to distinguish different devices, modules, or units, and are not used to limit the order or interdependence relationship of the functions performed by these devices, modules, or units.

[0018] It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0019] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only for illustrative purposes and are not used to limit the scope of these messages or information.

[0020] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0021] Referring to Figure 1 , a flowchart 100 of some embodiments of a fault detection method applied to a distribution network according to the present disclosure is shown. The fault detection method applied to the distribution network includes the following steps:

[0022] Step 101, locate at least one node to be detected in the distribution network to obtain a set of node information to be detected.

[0023] In some embodiments, an execution subject (e.g., a computing device) of the fault detection method applied to the distribution network may locate at least one node to be detected in the distribution network to obtain a set of node information to be detected. Among them, the distribution network is used to distribute the electric energy transmitted by the transmission network to the power consumption terminals. Specifically, the distribution network may include, but is not limited to: low-voltage distribution network, medium-voltage distribution network, and high-voltage distribution network. The transmission network is a transmission network for transmitting the current generated by the power generation equipment. The power consumption terminal is a terminal that needs to consume current. For example, the power consumption terminal may be a factory or a power-consuming device. The node to be detected represents a vulnerable or damaged device that needs to be frequently inspected in the distribution network, or a device that affects the normal operation of the distribution network after performing corresponding operations. For example, the node to be detected may be an insulator, or a transformer, a circuit breaker, etc. involved in the grid loop closing operation. The node to be detected and the node information to be detected correspond one by one. The node information to be detected in the set of node information to be detected includes: the type of the node to be detected, the location of the node to be detected, and the local network topology diagram. The type of the node to be detected represents the device type of the device corresponding to the node to be detected. For example, the type of the node to be detected may be "transformer". The local network topology diagram represents the topological structure of the local distribution network where the node to be detected is located. In practice, the distribution network often has a mesh structure, and any device is connected to at least one other device through the distribution network. When a fault occurs in a certain device, it may affect the device stability of the devices connected thereto. Therefore, through the local network topology diagram, the connection relationship between the node to be detected when a fault occurs and the devices that may be affected can be represented. The local distribution network is a subnet of the above distribution network.

[0024] As an example, the above execution subject may traverse the above distribution network to determine the nodes to be detected in the distribution network, and pull the node information to be detected corresponding to the nodes to be detected to obtain the above set of node information to be detected.

[0025] It should be noted that the above computing device can be hardware or software. When the computing device is hardware, it can be implemented as a distributed cluster composed of multiple servers or terminal devices, or as a single server or a single terminal device. When the computing device is embodied as software, it can be installed in the above-listed hardware devices. It can be implemented as, for example, multiple software or software modules for providing distributed services, or as a single software or software module. No specific limitation is made here. It should be understood that the number of computing devices can be arbitrary according to implementation requirements.

[0026] In some optional implementation manners of some embodiments, the above execution subject locates at least one node to be inspected in the distribution network, and obtains a set of information on the nodes to be inspected, including:

[0027] In the first step, channel monitoring is performed on the control channel.

[0028] Among them, the above control channel is a bidirectional communication channel. In practice, the control channel can be a communication channel through which the execution subject sends control signals to the devices in the distribution network and receives information and / or data sent by the devices in the distribution network to the execution subject.

[0029] In the second step, in response to the timing reminder request initiated by the first device being monitored, the device information of the first device is pulled as the information of the node to be inspected in the set of information on the nodes to be inspected.

[0030] Among them, the first device is the device in the above distribution network corresponding to the end of the inspection timer. The inspection timer is a timer used to prompt the inspection of the device. In practice, each device in the distribution network will be set with a corresponding inspection timer. When the inspection timer reaches zero, it will trigger a corresponding timing reminder request. When the execution entity monitors the timing reminder request through the control channel, the device corresponding to the end of the inspection timer of the corresponding inspection timer will be used as the first device. At the same time, the device information corresponding to the first device is pulled up as the information of the node to be inspected. For example, the pulled device information may include: device type, device location, and the network topology diagram of the local network where the device is located. Among them, the inspection timer is controlled by the device type, historical inspection records, and the number of failures of the device. Specifically, there are differences in the usage cycles of different devices. For example, the device A has high device stability, so the corresponding number of inspections and the number of frequent failures are less. At this time, a longer inspection duration can be set for the inspection timer of device A. Another example is that device B has poor device stability, so the corresponding number of inspections and the number of failures are more. Therefore, a shorter inspection duration can be set for the inspection timer of device B. Specifically, a basic duration corresponding to the device type can be set, and on the basis of the basic duration, the basic duration is scaled by the number of inspections and the number of failures corresponding to the historical inspection records as the timing duration of the inspection timer. Among them, the timing duration of the inspection timer is inversely proportional to the number of inspections and the number of failures corresponding to the historical inspection records, that is, the more the number of inspections or the number of failures corresponding to the historical inspection records, the shorter the timing duration of the inspection timer.

[0031] In the third step, in response to monitoring the target device operation for the second device, pull up the device information of the second device as the information of the node to be inspected in the set of information of the nodes to be inspected.

[0032] Among them, the second device is the device in the above distribution network for closing the loop of the distribution network, and the target device operation is the operation of closing the loop of the distribution network. For example, the second device may be a transformer.

[0033] Step 102: For each piece of information of the node to be inspected in the set of information of the nodes to be inspected, perform the following processing steps:

[0034] Step 1021: Collect real-time node status information according to the type of the node to be inspected and the location of the node to be inspected included in the information of the node to be inspected.

[0035] In some embodiments, the above-mentioned execution entity may collect real-time node status information according to the type of the node to be inspected and the location of the node to be inspected included in the information of the node to be inspected. Among them, the real-time node status information may characterize the real-time operating status corresponding to the node to be inspected. Specifically, it includes but is not limited to: the appearance image of the node to be inspected (device) and the device operation parameters. The device operation parameters may include but are not limited to: the bus voltage of the closed-loop feeder corresponding to the node to be inspected (device), the equivalent load of the superior power grid corresponding to the node to be inspected (device), the reactive power of the node to be inspected (device), the transmission power of the feeder corresponding to the node to be inspected (device), the active power of the node to be inspected (device), and the power flow of the superior power grid directly connected to the feeder corresponding to the node to be inspected (device). In practice, different device types require different real-time node status information to be collected. For example, when the node to be inspected is an insulator, since the insulator is an insulating control device arranged between the conductor and the grounding member, when the insulator is damaged, it will affect the operation of the distribution network. Therefore, the real-time node status information may be the appearance image corresponding to the insulator. Another example is that when the node to be inspected is a transformer, relevant parameters in the current transmission process need to be considered to avoid damage caused by the breakdown of the transformer. At this time, the real-time node status information may be the device operation parameters corresponding to the transformer.

[0036] In some optional implementation manners of some embodiments, the above-mentioned execution entity collects real-time node status information according to the type of the node to be inspected and the location of the node to be inspected included in the information of the node to be inspected, including:

[0037] First step, in response to the type of the node to be inspected included in the information of the node to be inspected being the device type corresponding to the device not involved in the closed-loop of the distribution network, determine the collection area according to the location of the node to be inspected included in the information of the node to be inspected.

[0038] Among them, the above-mentioned collection area is an electronic fence area centered on the location of the node to be inspected included in the information of the node to be inspected, and the radius of the collection area is controlled by the device density of the monitoring devices included in the collection area. In practice, the device density is inversely proportional to the area radius. Specifically, when the number of monitoring devices in the collection area is large, that is, the device density is high, the area radius of the collection area can be small at this time. When the number of monitoring devices in the collection area is small, that is, the device density is low, the area radius of the collection area can be large at this time. Therefore, the probability of collecting the appearance image (real-time device image) corresponding to the node to be inspected can be increased as much as possible.

[0039] Second step, determine the device operation status of each of at least one monitoring device in the above-mentioned collection area.

[0040] Among them, the device operation state characterizes the working state of the monitoring device. In practice, since the monitoring device needs to transmit the real-time collected monitoring video during the working state, the monitoring device is often in an online state. Therefore, the above-mentioned execution entity can determine the device operation state of each monitoring device in at least one monitoring device in the above-mentioned acquisition area through wired connection or wireless connection.

[0041] The third step is to determine whether there is a monitoring device that meets the screening conditions among the above-mentioned at least one monitoring device.

[0042] Among them, the above-mentioned screening conditions are: the corresponding device operation state characterizes that the monitoring device is running normally and the position of the node to be inspected included in the information of the node to be inspected is within the monitoring area of the monitoring device. In practice, each monitoring device corresponds to a monitoring angle. When the position of the monitoring device, the monitoring angle of the monitoring device, and the position of the node to be inspected of the node to be inspected are known, it can be calculated whether the node to be inspected is within the monitoring area of the monitoring device.

[0043] The fourth step is to, in response to the existence, control the monitoring device that meets the above-mentioned screening conditions to collect real-time device images.

[0044] Among them, the above-mentioned real-time device image is the appearance image of the node to be inspected corresponding to the above-mentioned information of the node to be inspected. In practice, the above-mentioned execution entity can control the monitoring device that meets the screening conditions to change the focal length to collect an image containing the device corresponding to the information of the node to be inspected as the real-time device image.

[0045] The fifth step is to, in response to the non-existence, plan a drone path according to the position of the node to be inspected included in the above-mentioned information of the node to be inspected.

[0046] Among them, the drone path is a flight path with the position of the node to be inspected included in the above-mentioned information of the node to be inspected as the end position and the drone take-off point as the start position. In practice, the above-mentioned execution entity plans the above-mentioned drone path in combination with a pre-collected high-precision map of the distribution network. Specifically, since the distribution network includes overhead lines, the drone has a problem of low accuracy in identifying overhead lines. Especially in an environment with insufficient light, based on the occurrence of installation phenomena, it will cause damage to the drone at least, and serious damage to the distribution network lines, short circuits, etc. Therefore, when planning the drone path, by combining the pre-collected high-precision map, it is assisted to plan the drone path to avoid the occurrence of collision situations.

[0047] The sixth step is to determine the target drone according to the above-mentioned drone path.

[0048] Among them, the above target drone is a drone in an idle state and can fly to and fro along the above drone path. In practice, to avoid the inability to return due to insufficient battery power of the drone, therefore, the drone path is planned first, and then the target drone that can fly to and fro along the drone path is selected in combination with the drone path.

[0049] Step 7: Control the above target drone to collect real-time device images along the above drone path.

[0050] In practice, the above execution entity can send the above drone path to the target drone so that the target drone flies along the above drone path. At the same time, the drone can also adaptively fine-tune the drone path in combination with its own flight control system and environment detection system. When the target drone approaches the position of the node to be inspected, the target drone can control the camera to take pictures of the position of the node to be inspected to obtain the above real-time device images.

[0051] Step 8: Determine the above real-time device image as the above real-time node status information.

[0052] In practice, real-time device images from multiple angles can be collected as the above real-time node status information.

[0053] Step 9: In response to the device type corresponding to the node to be inspected included in the above node information to be inspected being the device type related to the closed-loop operation of the distribution network, collect the closed-loop current characteristics of the node to be inspected corresponding to the above node information to be inspected as the above real-time node status information.

[0054] Among them, the closed-loop current characteristics include but are not limited to: the bus voltage of the closed-loop feeder corresponding to the node to be inspected (device), the equivalent load of the superior power grid corresponding to the node to be inspected (device), the reactive power of the node to be inspected (device), the transmission power of the feeder corresponding to the node to be inspected (device), the active power of the node to be inspected (device), and the power flow of the superior power grid directly connected to the feeder corresponding to the node to be inspected (device).

[0055] Step 1022: Perform fault detection on the node to be inspected corresponding to the node information to be inspected according to the real-time node status information and the local network topology diagram included in the node information to be inspected, so as to generate node detection information.

[0056] In some embodiments, the above-mentioned execution entity may perform a fault detection on the to-be-detected node corresponding to the to-be-detected node information according to the real-time node status information and the local network topology map included in the to-be-detected node information, so as to generate node detection information. Among them, the above-mentioned node detection information includes: node fault type, node fault probability, and node impact degree. Among them, the node fault probability represents the occurrence probability of the fault corresponding to the node fault type. The node impact degree represents the influence degree of the to-be-detected node on the devices in the local distribution network corresponding to the local network topology map. The node fault type represents the fault type that occurs in the device corresponding to the to-be-detected node.

[0057] In some optional implementation manners of some embodiments, the above-mentioned execution entity performs a fault detection on the to-be-detected node corresponding to the to-be-detected node information according to the above-mentioned real-time node status information and the local network topology map included in the to-be-detected node information, so as to generate node detection information, including:

[0058] In the first step, in response to the above-mentioned real-time node status information being a real-time device image, perform an adaptive scaling on the above-mentioned real-time node status information to obtain a scaled device image.

[0059] In practice, the above-mentioned execution entity may perform operations such as scaling, cropping, splicing, and filling on the real-time node status information to obtain the above-mentioned scaled device image.

[0060] As an example, since the model numbers of different monitoring devices are different, there may be differences in the image sizes of the collected implementation device images. Therefore, it is necessary to perform a unified image size scaling on the real-time node information as the scaled device image. Among them, the image size of the scaled device image is 640×640. Specifically, when non-uniform scaling cannot be performed, for example, the image size of the real-time device image is A×B. Among them, A > B. At this time, the above-mentioned execution entity may perform scaling based on A, and at this time, fill 0 on both sides corresponding to B so that the image size of the scaled device image is 640×640.

[0061] In the second step, perform image enhancement on the above-mentioned scaled device image to obtain an enhanced device image.

[0062] In practice, the above-mentioned execution entity may perform Mosaic data enhancement on the scaled device image to obtain an enhanced device image.

[0063] In the third step, determine a local image through a pre-trained fault location model and the above-mentioned enhanced device image.

[0064] Among them, the above-mentioned local image is a local image at the position of the region of interest included in the above-mentioned enhanced device image. In practice, the fault location model includes: a first backbone network and a second backbone network.

[0065] Among them, the first backbone network includes: Convolutional Layer A, Convolutional Layer B, Bottleneck CSP Layer A, Convolutional Layer C, Bottleneck CSP Layer B, Convolutional Layer D, Bottleneck CSP Layer C, Convolutional Layer E, and SPPF Layer. Among them, the output of Convolutional Layer A is the input of Convolutional Layer B. The output of Convolutional Layer B is the input of Bottleneck CSP Layer A. The output of Bottleneck CSP Layer A is the input of Convolutional Layer C. The output of Convolutional Layer C is the input of Bottleneck CSP Layer B. The output of Bottleneck CSP Layer B is the input of Convolutional Layer D. The output of Convolutional Layer D is the input of Bottleneck CSP Layer C. The output of Bottleneck CSP Layer C is the input of Convolutional Layer E. The output of Convolutional Layer E is the input of SPPF Layer.

[0066] Specifically, the input channel number of Convolutional Layer A is 3, the output channel number is 64, the convolutional kernel size is 6×6, and the stride is 2. The input channel number of Convolutional Layer B is 64, the output channel number is 128, the convolutional kernel size is 3×3, and the stride is 2. The input and output channel numbers of Bottleneck CSP Layer A are both 128. The input channel number of Convolutional Layer C is 128, the output channel number is 256, the convolutional kernel size is 3×3, and the stride is 2. The input and output channel numbers of Bottleneck CSP Layer B are both 256. The input channel number of Convolutional Layer D is 256, the output channel number is 512, the convolutional kernel size is 3×3, and the stride is 2. The input and output channel numbers of Bottleneck CSP Layer C are both 512. The input channel number of Convolutional Layer E is 512, the output channel number is 1024, the convolutional kernel size is 3×3, and the stride is 2. The output channel number of SPPF Layer is 1024.

[0067] Among them, the second backbone network includes: Bottleneck CSP layer D, convolutional layer F, upsampling layer A, concatenation layer A, Bottleneck CSP layer E, convolutional layer G, upsampling layer B, concatenation layer B, Bottleneck CSP layer F, convolutional layer H, concatenation layer C, Bottleneck CSP layer G, convolutional layer I, concatenation layer D, and Bottleneck CSP layer H. The input of Bottleneck CSP layer D is the SPPF layer. The output of Bottleneck CSP layer D is the input of convolutional layer F. The output of convolutional layer F is the input of upsampling layer A and the input of concatenation layer D. The output of upsampling layer A is the input of concatenation layer A, and at the same time, the output of Bottleneck CSP layer C is also the input of concatenation layer A. The output of concatenation layer A is the input of Bottleneck CSP layer E. The output of Bottleneck CSP layer E is the input of convolutional layer G. The output of convolutional layer G is the input of upsampling layer B, and at the same time, the output of convolutional layer G is also the input of concatenation layer C. The output of upsampling layer B is the input of concatenation layer B, and at the same time, the output of Bottleneck CSP layer B is also the input of concatenation layer B. The output of concatenation layer B is the output of Bottleneck CSP layer F. The output of Bottleneck CSP layer F is the input of convolutional layer H. The output of convolutional layer H is the input of concatenation layer C. The output of concatenation layer C is the input of Bottleneck CSP layer G. The output of Bottleneck CSP layer G is the input of convolutional layer I. The output of convolutional layer I is the input of concatenation layer D. The output of concatenation layer D is the input of Bottleneck CSP layer H. Among them, the outputs of Bottleneck CSP layer H, Bottleneck CSP layer G, and Bottleneck CSP layer F are local images.

[0068] Specifically, the number of input channels and output channels of Bottleneck CSP layer D are both 1024. The number of input channels of convolutional layer F is 1024, the number of output channels is 512, the convolutional kernel size is 1×1, and the stride is 1. The number of input channels and output channels of upsampling layer A are both 512. The number of input channels and output channels of concatenation layer A are both 1024. The number of input channels of Bottleneck CSP layer E is 1024, and the number of output channels is 512. The number of input channels of convolutional layer G is 512, the number of output channels is 256, the convolutional kernel size is 1×1, and the stride is 1. The number of input channels and output channels of upsampling layer B are both 256. The number of input channels and output channels of concatenation layer B are both 512. The number of input channels of Bottleneck CSP layer F is 512, and the number of output channels is 256. The number of input channels of convolutional layer H is 256, the number of output channels is 256, the convolutional kernel size is 3×3, and the stride is 2. The number of input channels and output channels of concatenation layer C are both 512. The number of input channels and output channels of Bottleneck CSP layer G are both 512. The number of input channels of convolutional layer I is 512, the number of output channels is 512, the convolutional kernel size is 3×3, and the stride is 2. The number of input channels and output channels of concatenation layer D are both 1024. The number of input channels and output channels of Bottleneck CSP layer H are both 1024. By adopting the SPPF layer, that is, serially passing the input through multiple maxpool layers included in the SPPF layer, the calculation time can be significantly shortened, and at the same time, the purpose of compressing the model can also be achieved. In addition, through the feature pyramid structure, the second backbone network can effectively transfer semantic features and strengthen the localization information from top to bottom.

[0069] In addition, the above fault location model adopts the CIOU_Loss loss function to calculate the intersection over union loss, and filters the regions of interest through non-maximum suppression. In addition, considering the problem of device occlusion, weighted processing is performed during non-maximum suppression to filter out overlapping regions of interest. In addition, during non-maximum suppression, the intersection over union adopts the DIOU intersection over union to better detect overlapping devices.

[0070] In the fourth step, the above node detection information is generated through the pre-trained fault classification model, the above local image, and the local network topology map included in the above node information to be detected.

[0071] Among them, the above-mentioned fault location model and the above-mentioned fault classification model are included in the fault detection model, and the above-mentioned fault location model and the above-mentioned fault classification model are decoupled. In practice, considering that the monitoring device or the drone also has a certain computing power, therefore, the fault location model and / or the fault classification model can be pre-deployed on the monitoring device or the drone with sufficient computing power, so that after the monitoring device or the drone collects the real-time device image, it can directly generate the node detection information locally. In this way, only a certain amount of network traffic is consumed to transmit the node detection information calculated by the monitoring device or the drone back to the execution entity. This method can greatly improve the computing power utilization efficiency of the execution entity and the edge side (drone or monitoring device), and at the same time reduce the computing power pressure on the execution entity.

[0072] Optionally, the fault classification model includes: a first convolutional layer, a first feature reuse layer, a second convolutional layer, a first pooling layer, a second feature reuse layer, a third convolutional layer, a second pooling layer, a third feature reuse layer, a third pooling layer, and a linear layer.

[0073] Optionally, the above-mentioned execution entity generates the above-mentioned node detection information through the pre-trained fault classification model, the above-mentioned local image, and the local network topology map included in the above-mentioned node information to be detected, including:

[0074] Step 1: Perform convolutional processing on the above-mentioned local image through the above-mentioned first convolutional layer to obtain a first convolutional feature.

[0075] Step 2: Input the above-mentioned first convolutional feature into the above-mentioned first feature reuse layer to obtain a second convolutional feature.

[0076] Step 3: Perform convolutional processing on the above-mentioned second convolutional feature through the above-mentioned second convolutional layer to obtain a third convolutional feature.

[0077] Step 4: Perform pooling processing on the above-mentioned third convolutional feature through the above-mentioned first pooling layer to obtain a pooled third convolutional feature.

[0078] Step 5: Input the above-mentioned pooled third convolutional feature into the above-mentioned second feature reuse layer to obtain a fourth convolutional feature.

[0079] Step 6: Perform convolutional processing on the above-mentioned fourth convolutional feature through the above-mentioned third convolutional layer to obtain a fifth convolutional feature.

[0080] Step 7: Perform pooling processing on the above-mentioned fifth convolutional feature through the above-mentioned second pooling layer to obtain a pooled fifth convolutional feature.

[0081] Step 8: Input the above-mentioned pooled fifth convolutional feature into the above-mentioned third feature reuse layer to obtain a pooled fifth convolutional feature.

[0082] Step 9: Input the fifth convolutional feature after the above pooling into the above linear layer to obtain the node fault type and node fault probability included in the above node detection information.

[0083] Step 10: Determine the node spread degree included in the above node detection information according to the node fault type and node fault probability included in the above node detection information, and the local network topology map included in the above node information to be detected.

[0084] In practice, different node fault types correspond to different basic spread degrees. The node fault probability can be used as the weight A of the basic spread degree. In addition, the device density included in the local network topology map corresponds to the weight B. Specifically, the node spread degree = the basic spread degree corresponding to the node fault type × (weight A + weight B). Among them, the weights A and B are normalized before use so that the weights A and B are in the same order of magnitude.

[0085] Fifth step, in response to the above real-time node status information being the closed-loop current feature, perform feature preprocessing on the above real-time node status information to generate the preprocessed closed-loop current feature.

[0086] Among them, the closed-loop current feature includes but is not limited to: the bus voltage of the closed-loop feeder corresponding to the node (device) to be detected, the equivalent load of the superior power grid corresponding to the node (device) to be detected, the reactive power of the node (device) to be detected, the transmission power of the feeder corresponding to the node (device) to be detected, the active power of the node (device) to be detected, and the power flow of the superior power grid directly connected to the feeder corresponding to the node (device) to be detected. In practice, the above execution entity can perform feature preprocessing on the above real-time node status information by means such as outlier removal and data normalization to generate the preprocessed closed-loop current feature.

[0087] Sixth step, generate the above node detection information according to the above preprocessed closed-loop current feature, the pre-trained closed-loop current prediction model, and the local network topology map included in the above node information to be detected.

[0088] In practice, the above closed-loop current prediction model adopts the model structure of CNN (Convolutional Neural Networks, convolutional neural network) model + LSTM (Long Short-Term Memory, long short-term memory) model. Specifically, first use the CNN model to perform linear convolutional processing on the preprocessed closed-loop current feature, and then input the processed convolutional feature into the LSTM model. In addition, the LSTM model is connected with a fully connected layer for outputting the node fault type and node fault probability. Based on the node fault type and node fault probability, the method for calculating the node spread degree in step 10 of the above fourth step can be used to determine the node spread degree, which will not be elaborated here.

[0089] In addition, based on the output of the LTSM model, the above-mentioned execution entity can further calculate the allowable confluence current in a safe situation by using the method of quantile regression.

[0090] Step 1023: In response to the node failure probability being greater than the preset node failure probability and the node impact degree being less than the preset node impact degree, initiate a failure warning that matches the node failure type.

[0091] In some embodiments, the above-mentioned execution entity can initiate a failure warning that matches the node failure type in response to the node failure probability being greater than the preset node failure probability and the node impact degree being less than the preset node impact degree. In practice, the above-mentioned execution entity can initiate a failure warning that matches the node failure type for the device corresponding to the node to be inspected on the warning platform. In this case, only initiating the failure warning for the device corresponding to the node to be inspected can effectively avoid the problem of inaccurate subsequent fault location caused by large-scale alarms.

[0092] Step 1024: In response to the node failure probability being greater than the preset node failure probability and the node impact degree being greater than or equal to the preset node impact degree, initiate a risk warning for the local distribution network corresponding to the local network topology diagram included in the information of the node to be inspected.

[0093] In some embodiments, in response to the node failure probability being greater than the preset node failure probability and the node impact degree being greater than or equal to the preset node impact degree, initiate a risk warning for the local distribution network corresponding to the local network topology diagram included in the information of the node to be inspected. In practice, the above-mentioned execution entity can initiate a risk warning for the local distribution network corresponding to the local network topology diagram included in the information of the node to be inspected on the warning platform. In this case, effective failure warnings can be issued for possible cascading failures, avoiding the impact on the overall operation of the distribution network caused by large-area cascading failures.

[0094] The above-mentioned various embodiments of the present disclosure have the following beneficial effects: Through the fault detection method applied to the distribution network in some embodiments of the present disclosure, effective fault detection for the distribution network is achieved, ensuring the stability of the distribution network, reducing the risk of equipment damage, and reducing the losses caused by power supply interruption. Specifically, the reasons for faults in the distribution network are as follows: The structure of large-scale distribution networks is complex. When a device in the distribution network fails, it is extremely likely to affect the overall stability of the distribution network, thereby affecting other devices and causing equipment damage and power supply interruption. Based on this, in some embodiments of the present disclosure, the fault detection method applied to the distribution network first locates at least one node to be detected in the distribution network to obtain a set of information about the nodes to be detected. Among them, the above-mentioned distribution network is used to distribute the electric energy transmitted by the transmission network to the power consumption terminals, and the nodes to be detected correspond one-to-one with the information about the nodes to be detected. The information about the nodes to be detected in the above-mentioned set of information about the nodes to be detected includes: the type of the node to be detected, the location of the node to be detected, and a local network topology diagram, where the local network topology diagram characterizes the topological structure of the local distribution network where the node to be detected is located, and the local distribution network is a subnet of the above-mentioned distribution network. In practice, the network structure of large-scale distribution networks is complex. If the entire distribution network is subjected to fault detection, the detection difficulty is large and the detection period is long. Therefore, by locating the key nodes (nodes to be detected) in the distribution network, the detection complexity can be reduced. Secondly, for each piece of information about the nodes to be detected in the above-mentioned set of information about the nodes to be detected, the following processing steps are executed: First step, according to the type of the node to be detected and the location of the node to be detected included in the above-mentioned information about the nodes to be detected, real-time node status information is collected. By collecting the real-time node status, the current operating status of the device corresponding to the node to be detected can be obtained. Second step, based on the above-mentioned real-time node status information and the local network topology diagram included in the above-mentioned information about the nodes to be detected, fault detection is performed on the node to be detected corresponding to the above-mentioned information about the nodes to be detected to generate node detection information, where the above-mentioned node detection information includes: the type of node fault, the probability of node fault, and the degree of node influence, where the probability of node fault represents the occurrence probability of the fault corresponding to the type of node fault, and the degree of node influence represents the influence degree of the node to be detected on the devices in the local distribution network corresponding to the local network topology diagram. In this way, from the perspective of the node to be detected itself and the influence of the node to be detected on the surrounding devices, the influence of the type of node fault on the surrounding devices is detected. Third step, in response to the above-mentioned probability of node fault being greater than the preset probability of node fault and the above-mentioned degree of node influence being less than the preset degree of node influence, a fault warning matching the above-mentioned type of node fault is initiated. In this way, single-point warning for the device corresponding to the node to be detected is realized. Fourth step, in response to the above-mentioned probability of node fault being greater than the above-mentioned preset probability of node fault and the above-mentioned degree of node influence being greater than or equal to the above-mentioned preset degree of node influence, a risk warning for the local distribution network corresponding to the local network topology diagram included in the above-mentioned information about the nodes to be detected is initiated. In this way, overall warning for the node to be detected and the local distribution network where it is located is realized.In this way, fault detection can be efficiently achieved, ensuring the stability of the distribution network, reducing the risk of equipment damage, and reducing the losses caused by power supply interruptions.

[0095] Further reference is made to Figure 2 , as an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of a fault detection device applied to a distribution network. These device embodiments correspond to Figure 1 the method embodiments shown, and the fault detection device applied to the distribution network can be specifically applied to various electronic devices.

[0096] As Figure 2 shown, some embodiments of the fault detection device 200 applied to a distribution network include: a positioning unit 201 and an execution unit 202. Among them, the positioning unit 201 is configured to locate at least one node to be detected in the distribution network to obtain a set of information about the nodes to be detected. Among them, the above distribution network is used to distribute the electric energy transmitted by the transmission network to the power consumption terminals. The nodes to be detected and the information about the nodes to be detected are in one-to-one correspondence. The information about the nodes to be detected in the above set of information about the nodes to be detected includes: the type of the node to be detected, the location of the node to be detected, and a local network topology diagram. Among them, the local network topology diagram characterizes the topological structure of the local distribution network where the node to be detected is located, and the local distribution network is a subnet of the above distribution network; the execution unit 202 is configured to perform the following processing steps for each piece of information about the nodes to be detected in the above set of information about the nodes to be detected: collect real-time node status information according to the type of the node to be detected and the location of the node to be detected included in the above information about the nodes to be detected; perform fault detection on the node to be detected corresponding to the above information about the nodes to be detected according to the above real-time node status information and the local network topology diagram included in the above information about the nodes to be detected to generate node detection information. Among them, the above node detection information includes: the type of node fault, the probability of node fault, and the degree of node influence. Among them, the probability of node fault characterizes the occurrence probability of the fault corresponding to the type of node fault, and the degree of node influence characterizes the influence degree of the node to be detected on the equipment in the local distribution network corresponding to the local network topology diagram; in response to the above probability of node fault being greater than a preset probability of node fault and the above degree of node influence being less than a preset degree of node influence, initiate a fault warning matching the above type of node fault; in response to the above probability of node fault being greater than the above preset probability of node fault and the above degree of node influence being greater than or equal to the above preset degree of node influence, initiate a risk warning for the local distribution network corresponding to the local network topology diagram included in the above information about the nodes to be detected.

[0097] It can be understood that the various units described in the fault detection device 200 applied to the distribution network correspond to the respective steps in the method described with reference to Figure 1 Therefore, the operations, features, and beneficial effects described above for the method also apply to the fault detection device 200 applied to the distribution network and the units included therein, and will not be repeated here.

[0098] Reference is made below to Figure 3 , which shows a schematic structural diagram of an electronic device (e.g., a computing device) 300 suitable for implementing some embodiments of the present disclosure. Figure 3 The illustrated electronic device is merely an example and should not impose any limitations on the functions and usage scope of the embodiments of the present disclosure.

[0099] As Figure 3 shown, the electronic device 300 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which may perform various appropriate actions and processes according to a program stored in the read-only memory 302 or a program loaded from the storage device 308 into the random access memory 303. In the random access memory 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the read-only memory 302, and the random access memory 303 are connected to each other through a bus 304. The input / output interface 305 is also connected to the bus 304.

[0100] Generally, the following devices may be connected to the input / output interface 305: an input device 306 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 308 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 309. The communication device 309 may allow the electronic device 300 to communicate with other devices wirelessly or wiredly to exchange data. Although Figure 3 the electronic device 300 with various devices is shown, it should be understood that it is not required to implement or include all the shown devices. Instead, more or fewer devices may be implemented or included. Figure 3 Each block shown in

[0101] Specifically, according to some embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, some embodiments of the present disclosure include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes program codes for performing the methods shown in the flowcharts. In such some embodiments, the computer program may be downloaded and installed from the network through the communication device 309, or installed from the storage device 308, or installed from the read-only memory 302. When the computer program is executed by the processing device 301, the above-mentioned functions defined in the methods of some embodiments of the present disclosure are performed.

[0102] It should be noted that the computer-readable media described in some embodiments of the present disclosure may be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of the present disclosure, the computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium may also be any computer-readable medium other than the computer-readable storage medium, and the computer-readable signal medium may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted by any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0103] In some embodiments, the client and the server may communicate using any currently known or future-developed network protocol such as HTTP (Hyper Text Transfer Protocol), and may be interconnected with digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include local area networks ("LANs"), wide area networks ("WANs"), the Internet (e.g., the Internet), and end-to-end networks (e.g., ad hoc end-to-end networks), as well as any currently known or future-developed networks.

[0104] The above computer-readable medium may be included in the above electronic device; or may exist independently without being assembled into the electronic device. The above computer-readable medium carries one or more programs. When the above one or more programs are executed by the electronic device, the electronic device is caused to: locate at least one node to be inspected in the distribution network to obtain a set of node information to be inspected, wherein the distribution network is used to distribute the electric energy transmitted by the transmission network to the power consumption terminals, the nodes to be inspected and the node information to be inspected are in one-to-one correspondence, and the node information to be inspected in the set of node information to be inspected includes: the type of the node to be inspected, the location of the node to be inspected, and a local network topology diagram, wherein the local network topology diagram characterizes the topological structure of the local distribution network where the node to be inspected is located, and the local distribution network is a subnet of the above distribution network; for each piece of node information to be inspected in the set of node information to be inspected, perform the following processing steps: collect real-time node status information according to the type of the node to be inspected and the location of the node to be inspected included in the above node information to be inspected; perform fault detection on the node to be inspected corresponding to the above node information to be inspected according to the above real-time node status information and the local network topology diagram included in the above node information to be inspected, so as to generate node detection information, wherein the node detection information includes: the type of node fault, the probability of node fault, and the degree of node impact, wherein the probability of node fault characterizes the occurrence probability of the fault corresponding to the type of node fault, and the degree of node impact characterizes the influence degree of the node to be inspected on the devices in the local distribution network corresponding to the local network topology diagram; in response to the probability of node fault being greater than a preset probability of node fault and the degree of node impact being less than a preset degree of node impact, initiate a fault warning matching the type of node fault; in response to the probability of node fault being greater than the preset probability of node fault and the degree of node impact being greater than or equal to the preset degree of node impact, initiate a risk warning for the local distribution network corresponding to the local network topology diagram included in the above node information to be inspected.

[0105] Computer program code for performing the operations of some embodiments of the present disclosure may be written in one or more programming languages or combinations thereof. The above programming languages include object-oriented programming languages - such as Java, Smalltalk, C++; and also include conventional procedural programming languages - such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network - including a local area network (LAN) or a wide area network (WAN) - or may be connected to an external computer (for example, by using an Internet service provider to connect through the Internet).

[0106] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram may represent a module, a segment of a program, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, as well as combinations of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0107] The units described in some embodiments of the present disclosure can be implemented in software or in hardware. The described units can also be provided in a processor. For example, it can be described as: a processor includes a positioning unit and an execution unit. Among them, the names of these units do not constitute a limitation to the unit itself in some cases. For example, the positioning unit can also be described as "positioning at least one node to be inspected in a distribution network to obtain a set of information of the nodes to be inspected, where the above distribution network is used to distribute the electric energy transmitted by the transmission network to the power consumption terminals, the nodes to be inspected and the information of the nodes to be inspected are in one-to-one correspondence, and the information of the nodes to be inspected in the above set of information of the nodes to be inspected includes: the type of the node to be inspected, the location of the node to be inspected, and a local network topology diagram, where the local network topology diagram represents the topology structure of the local distribution network where the node to be inspected is located, and the local distribution network is a subnet of the above distribution network".

[0108] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that can be used include: Field Programmable Gate Arrays (FPGA), Application Specific Integrated Circuits (ASIC), Application Specific Standard Products (ASSP), System on Chip (SOC), Complex Programmable Logic Devices (CPLD), and so on.

[0109] The above description is only some preferred embodiments of the present disclosure and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the technical solutions formed by mutually replacing the above features with the technical features (but not limited to) disclosed in the embodiments of the present disclosure that have similar functions.

Claims

1. A fault detection method applied to a distribution network, comprising: Locate at least one node to be inspected in the distribution network, and obtain a set of information about the node to be inspected, wherein the distribution network is used to distribute the electric energy transmitted by the transmission network to the power consumption terminal, and the node to be inspected corresponds to the node to be inspected information one by one, and the node to be inspected information in the set of information about the node to be inspected includes: the type of the node to be inspected, the location of the node to be inspected, and a local network topology map, wherein the local network topology map represents the topological structure of the local distribution network where the node to be inspected is located, and the local distribution network is a subnet of the distribution network; For each piece of node information to be inspected in the set of node information to be inspected, the following processing steps are performed: Collecting real-time node status information according to the node type and node location included in the node information; According to the real-time node status information and the local network topology included in the node information to be inspected, fault detection is performed on the node to be inspected corresponding to the node information to be inspected to generate node detection information, wherein the node detection information includes: node fault type, node fault probability and node spread, wherein the node fault probability represents the probability of occurrence of the fault corresponding to the node fault type, and the node spread represents the influence of the node to be inspected on the equipment in the local distribution network corresponding to the local network topology. According to the real-time node status information and the local network topology included in the node information to be inspected, fault detection is performed on the node to be inspected corresponding to the node information to be inspected to generate node detection information, including: in response to the real-time node status information being a real-time device image, a local image is determined through a pre-trained fault location model and the real-time device; the node detection information is generated through a pre-trained fault classification model, the local image and the local network topology included in the node information to be inspected; in response to the real-time node status information being a closed-loop current feature, the node detection information is generated according to the pre-processed closed-loop current feature, the pre-trained closed-loop current prediction model and the local network topology included in the node information to be inspected; In response to the node failure probability being greater than a preset node failure probability and the node spread being less than a preset node spread, initiating a fault warning matching the node failure type; In response to the node failure probability being greater than the preset node failure probability and the node reach being greater than or equal to the preset node reach, a risk warning for a local distribution network corresponding to a local network topology diagram included in the node information to be inspected is initiated.

2. The method according to claim 1, wherein: The step of locating at least one node to be inspected in the power distribution network and obtaining a set of information of the node to be inspected includes: Performing channel monitoring on a control channel, wherein the control channel is a bidirectional communication channel; In response to monitoring a timing reminder request initiated by a first device, pulling device information of the first device as the node information to be inspected in the node information set to be inspected, wherein the first device is a device in the distribution network and the timing of the corresponding inspection timer has expired, and the inspection timer is controlled by the device type, historical inspection records and number of faults of the device; In response to monitoring a target device operation for a second device, the device information of the second device is pulled as the node information to be inspected in the node information set to be inspected, wherein the second device is a device within the distribution network used for distribution network closing, and the target device operation is a distribution network closing operation.

3. The method according to claim 2, wherein: The collecting of real-time node status information according to the node type and node position included in the node information includes: In response to the type of the node to be inspected included in the node to be inspected information being a device type corresponding to a device not involved in the power distribution network loop closure, a collection area is determined according to the position of the node to be inspected included in the node to be inspected information, wherein the collection area is an electronic fence area centered on the position of the node to be inspected included in the node to be inspected information, and the area radius of the collection area is controlled by the device density of the monitoring devices contained in the collection area; Determining a device operating status of each monitoring device in at least one monitoring device within the collection area; Determine whether there is a monitoring device that meets the screening condition in the at least one monitoring device, wherein the screening condition is: the corresponding device operation status indicates that the monitoring device is operating normally and the position of the node to be inspected included in the node to be inspected information is located within the monitoring area of ​​the monitoring device; In response to existence, controlling the monitoring device that meets the screening condition to collect a real-time device image, wherein the real-time device image is an appearance image of the node to be inspected corresponding to the information of the node to be inspected; In response to the absence of the node, planning a path of the drone according to the location of the node to be inspected included in the information of the node to be inspected; Determine a target drone according to the drone path, wherein the target drone is a drone that is in an idle state and can travel back and forth along the drone path; Controlling the target UAV along the UAV path to collect real-time device images; Determine the real-time device image as the real-time node status information; In response to the node type to be inspected included in the node information to be inspected being a device type corresponding to a device involved in the closing loop of a distribution network, closing loop current characteristics of the node to be inspected corresponding to the node information to be inspected are collected as the real-time node status information.

4. The method according to claim 3, wherein: The performing fault detection on the node to be inspected corresponding to the node to be inspected information according to the real-time node status information and the local network topology diagram included in the node to be inspected information to generate node detection information includes: In response to the real-time node status information being a real-time device image, adaptively scaling the real-time node status information to obtain a scaled device image; Performing image enhancement on the scaled device image to obtain an enhanced device image; Determine a local image by using a pre-trained fault location model and the enhanced device image, wherein the local image is a local image of a location where an area of ​​interest is located and is contained in the enhanced device image; Generate the node detection information through a pre-trained fault classification model, the local image and a local network topology map included in the information of the node to be detected, wherein the fault location model and the fault classification model are included in the fault detection model, and the fault location model and the fault classification model are decoupled; In response to the real-time node status information being a closed-loop current feature, performing feature preprocessing on the real-time node status information to generate a preprocessed closed-loop current feature; The node detection information is generated according to the pre-processed closed-loop current characteristics, the pre-trained closed-loop current prediction model and the local network topology diagram included in the node information to be detected.

5. The method according to claim 4, wherein: The fault classification model includes: a first convolution layer, a first feature multiplexing layer, a second convolution layer, a first pooling layer, a second feature multiplexing layer, a third convolution layer, a second pooling layer, a third feature multiplexing layer, a third pooling layer and a linear layer; and The node detection information is generated by using the pre-trained fault classification model, the local image and the local network topology map included in the node information to be detected, including: Performing convolution processing on the local image through the first convolution layer to obtain a first convolution feature; Inputting the first convolutional feature into the first feature multiplexing layer to obtain a second convolutional feature; Performing convolution processing on the second convolution feature through the second convolution layer to obtain a third convolution feature; Performing pooling processing on the third convolutional feature through the first pooling layer to obtain a pooled third convolutional feature; Inputting the pooled third convolutional feature into the second feature multiplexing layer to obtain a fourth convolutional feature; Performing convolution processing on the fourth convolution feature through the third convolution layer to obtain a fifth convolution feature; Performing pooling processing on the fifth convolutional feature through the second pooling layer to obtain a pooled fifth convolutional feature; Inputting the pooled fifth convolutional feature into the third feature multiplexing layer to obtain the pooled fifth convolutional feature; Inputting the pooled fifth convolutional feature into the linear layer to obtain the node fault type and node fault probability included in the node detection information; The node impact degree included in the node detection information is determined according to the node failure type and node failure probability included in the node detection information and the local network topology map included in the to-be-detected node information.

6. A fault detection device for a power distribution network, comprising: A positioning unit is configured to locate at least one node to be inspected in the distribution network, and obtain a set of information about the node to be inspected, wherein the distribution network is used to distribute the electric energy transmitted by the transmission network to the power consumption terminal, and the node to be inspected corresponds to the node to be inspected information one by one, and the node to be inspected information in the set of information about the node to be inspected includes: the type of the node to be inspected, the location of the node to be inspected, and a local network topology map, wherein the local network topology map represents the topology structure of the local distribution network where the node to be inspected is located, and the local distribution network is a subnet of the distribution network; The execution unit is configured to perform the following processing steps for each node information to be inspected in the node information set to be inspected: collecting real-time node status information according to the node type to be inspected and the node position to be inspected included in the node information to be inspected; performing fault detection on the node to be inspected corresponding to the node information to be inspected according to the real-time node status information and the local network topology map included in the node information to be inspected, so as to generate node detection information, wherein the node detection information includes: node fault type, node fault probability and node spread, wherein the node fault probability represents the probability of occurrence of the fault corresponding to the node fault type, and the node spread represents the influence of the node to be inspected on the equipment in the local distribution network corresponding to the local network topology map, wherein performing fault detection on the node to be inspected corresponding to the node information to be inspected according to the real-time node status information and the local network topology map included in the node information to be inspected, so as to generate node detection information, includes: in response to The real-time node status information is a real-time device image. The local image is determined by a pre-trained fault location model and the real-time device; the node detection information is generated by a pre-trained fault classification model, the local image and the local network topology included in the node information to be inspected; in response to the real-time node status information being a closed-loop current feature, the node detection information is generated according to the pre-processed closed-loop current feature, the pre-trained closed-loop current prediction model and the local network topology included in the node information to be inspected; in response to the node failure probability being greater than a preset node failure probability and the node impact being less than a preset node impact, a fault warning matching the node failure type is initiated; in response to the node failure probability being greater than the preset node failure probability and the node impact being greater than or equal to the preset node impact, a risk warning for the local distribution network corresponding to the local network topology included in the node information to be inspected is initiated.

7. An electronic device comprising: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 5.

8. A computer readable medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images