Method and device for detecting opening and closing states of disconnecting switch and electronic equipment

Abstract

The invention provides a method and a device for detecting opening and closing states of an isolating switch, and electronic equipment. The method comprises the following steps: acquiring image data of the disconnecting switch; the image data of the disconnecting switch are analyzed through a target detection network, a conductive arm detection frame and a static contact detection frame are generated, coordinates of a target position of the conductive arm detection frame are extracted to obtain first coordinates, coordinates of a target position of the static contact detection frame are extracted to obtain second coordinates, and the static contact detection frame at least comprises an image of a static contact; and calculating the distance between the first coordinate and the second coordinate, determining the disconnecting switch to be in an opening state when the distance is greater than a first preset threshold value, and determining the disconnecting switch to be in a closing state when the distance is less than or equal to a second preset threshold value. According to the invention, the problem that the opening and closing states of the disconnecting switch cannot be remotely identified in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the technical field of image recognition of power equipment, and in particular to a method, device, computer-readable storage medium and electronic device for detecting the opening and closing states of an isolating switch. Background Art

[0002] The isolating switch is an electrical device used to interrupt or isolate the power supply when the circuit is interrupted. Its main function is to cut off the power supply when the circuit needs to be repaired, inspected or in an emergency, to ensure the safety of the operator and prevent equipment damage or accidents. At present, most of the substations in my country are still old-fashioned substations, and have not yet completed the transformation from traditional substations to smart substations. Most of the inspection tasks and fault detection in substations still rely on manual work. When the isolating switch performs the opening and closing attitude detection, the staff usually needs to observe the opening and closing status of the isolating switch on site, and feedback the results to the operator of the monitoring center before taking the next step. This operation method is not simple, has low safety, and is easily affected by the subjective judgment of the staff. With the development of substations towards intelligence, the opening and closing attitude detection of isolating switches has also developed from manual to automated, that is, the substation operator can directly and remotely visualize the opening and closing work of the isolating switch in the monitoring center. Therefore, adding the opening and closing attitude detection technology that can realize the true automation of the isolating switch is very meaningful for realizing truly unattended operation. Using the current advanced computer vision technology to identify the opening and closing status of substation disconnectors is a key step in the intelligentization of substations and is a problem that needs to be solved at present. Summary of the invention

[0003] The main purpose of the present application is to provide a method, device, computer-readable storage medium and electronic device for detecting the opening and closing states of an isolating switch, so as to at least solve the problem in the prior art that the opening and closing states of the isolating switch cannot be remotely identified.

[0004] To achieve the above-mentioned purpose, according to one aspect of the present application, a method for detecting the opening and closing state of an isolating switch is provided, comprising: acquiring isolating switch image data, wherein the isolating switch image data is image data at least including the isolating switch, and the isolating switch at least includes a conductive arm and a static contact; analyzing the isolating switch image data through a target detection network to generate a conductive arm detection frame and a static contact detection frame, and extracting the coordinates of the target position of the conductive arm detection frame to obtain a first coordinate, and extracting the coordinates of the target position of the static contact detection frame to obtain a second coordinate, wherein the conductive arm detection frame at least includes an image of the conductive arm, and the static contact detection frame at least includes an image of the static contact, and the target detection network is obtained by training multiple groups of data, and each group of data in the multiple groups of data includes historical image data of the isolating switch and a conductive arm detection frame label and a static contact detection frame label; calculating the distance between the first coordinate and the second coordinate, and determining that the isolating switch is in an opening state when the distance is greater than a first preset threshold, and determining that the isolating switch is in a closing state when the distance is less than or equal to a second preset threshold, wherein the first preset threshold is greater than the second preset threshold.

[0005] Optionally, before analyzing the isolating switch image data through the target detection network, the method also includes: acquiring the historical image data of the isolating switch; annotating the conductive arm of the isolating switch in the historical image data with a rotating frame to obtain a conductive arm detection frame, and annotating the vertex coordinates of the conductive arm detection frame to obtain the conductive arm detection frame label; annotating the static contact of the isolating switch with a horizontal frame to obtain a static contact detection frame, and annotating the vertex coordinates of the static contact detection frame to obtain the static contact detection frame label, wherein the rotating frame is obtained by rotating the horizontal frame; acquiring an initial target detection network, and training the initial target detection network with the historical image data, the conductive arm detection frame label and the static contact detection frame label to obtain the target detection network.

[0006] Optionally, the isolating switch image data is analyzed through a target detection network to generate a conductive arm detection frame and a static contact detection frame, including: extracting the appearance features, difference features and position features of the isolating switch through the target detection network, and performing feature fusion on the appearance features, difference features and position features to obtain a feature image after feature fusion, wherein the appearance features are features that characterize the appearance of the isolating switch, the difference features are features that characterize the differences between the isolating switches of different categories, and the position features are features that characterize the position changes of the isolating switch; obtaining the average size of the isolating switches in the isolating switch image data, and adjusting the size of the detection frame of the target detection network to the average size , generating a candidate area of ​​the feature image through the detection frame, wherein the detection frame is used to detect the conductive arm and the static contact, and the candidate area is a preliminarily determined area where the conductive arm and the static contact are located; extracting features of the candidate area to obtain candidate area features, extracting global space features and local space features from the candidate area features, fusing the global space features with the local space features, and updating the candidate area features, thereby generating the conductive arm detection frame and the static contact detection frame, wherein the candidate area features include all features in the candidate area, the global space features are features that characterize the conductive arm and the static contact, and the local space features are features that characterize the image background.

[0007] Optionally, extract global spatial features and local spatial features from the candidate region features, fuse the global spatial features and the local spatial features, and update the candidate region features, including: dividing the candidate region features into current candidate region features and other candidate region features; extracting the global spatial features through the current candidate region features, and extracting the local spatial features through other candidate region features; fusing the current candidate region features with the global spatial features to obtain a first fused feature, fusing the current candidate region features with the local spatial features to obtain a second fused feature, and fusing the first fused feature with the second fused feature to obtain the updated candidate region features.

[0008] Optionally, the global space feature is extracted through the feature of the current candidate area, including: obtaining an expansion coefficient, calculating extended vertex coordinates according to the vertex coordinates of the current candidate area and the expansion coefficient, obtaining the height and width of the isolating switch image, and updating the extended vertex coordinates according to the height and the width to obtain an extended candidate area; extracting the feature of the extended candidate area to obtain the global space feature.

[0009] Optionally, there are two static contacts, which are respectively located on both sides of the conductive arm, and the coordinates of the target position of the conductive arm detection frame are extracted to obtain the first coordinate, and the coordinates of the target position of the static contact detection frame are extracted to obtain the second coordinate, including: extracting the coordinates of the lower right corner position of the conductive arm detection frame to obtain the first coordinate; extracting the coordinates of the lower right corner position of the two static contact detection frames to obtain the second coordinate corresponding to each static contact detection frame diagram.

[0010] Optionally, the distance between the first coordinate and the second coordinate is calculated, and when the distance is greater than a first preset threshold, the disconnector is determined to be in the open state, and when the distance is less than or equal to a second preset threshold, the disconnector is determined to be in the closed state, including: respectively calculating the distance between the first coordinate and each of the second coordinates to obtain a first distance and a second distance; determining the minimum distance between the first distance and the second distance, and when the minimum distance is greater than the first preset threshold, determining that the disconnector is in the open state, and when the minimum distance is less than or equal to the second preset threshold, determining that the disconnector is in the closed state.

[0011] According to another aspect of the present application, a detection device for the opening and closing state of an isolating switch is provided, comprising: a first acquisition unit, used to acquire image data of the isolating switch, wherein the image data of the isolating switch is image data including at least the isolating switch, and the isolating switch includes at least a conductive arm and a static contact; an extraction unit, used to analyze the image data of the isolating switch through a target detection network, generate a conductive arm detection frame and a static contact detection frame, and extract the coordinates of the target position of the conductive arm detection frame to obtain a first coordinate, and extract the coordinates of the target position of the static contact detection frame to obtain a second coordinate, wherein the conductive arm detection frame is at least Including the image of the conductive arm, the static contact detection frame at least includes the image of the static contact, the target detection network is obtained by training multiple groups of data, each of the multiple groups of data includes historical image data of the disconnector and a conductive arm detection frame label and a static contact detection frame label; a determination unit is used to calculate the distance between the first coordinate and the second coordinate, and when the distance is greater than a first preset threshold, determine that the disconnector is in an open state, and when the distance is less than or equal to a second preset threshold, determine that the disconnector is in a closed state, wherein the first preset threshold is greater than the second preset threshold.

[0012] According to another aspect of the present application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the methods for detecting the opening and closing states of the isolating switch.

[0013] According to another aspect of the present application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing any one of the detection methods for the open and closed states of the isolating switch.

[0014] The technical solution of the present application is applied to obtain the isolating switch image data, which is the image data including at least the isolating switch, and the isolating switch includes at least the conductive arm and the static contact; the isolating switch image data is analyzed by the target detection network, the conductive arm detection frame and the static contact detection frame are generated, and the coordinates of the target position of the conductive arm detection frame are extracted to obtain the first coordinate, and the coordinates of the target position of the static contact detection frame are extracted to obtain the second coordinate, and the distance between the first coordinate and the second coordinate is calculated. When the distance is greater than the first preset threshold, the isolating switch is determined to be in the open state, and when the distance is less than or equal to the second preset threshold, the isolating switch is determined to be in the closed state. Compared with the prior art, which cannot remotely identify the opening and closing state of the isolating switch, the present application uses the target detection network to mark the conductive arm and the static contact of the isolating switch with a marking frame, and determines the opening and closing state of the isolating switch according to the distance of the coordinates of the marking frame, which can be remotely identified and judged, and the accuracy of identification and judgment is improved. Therefore, the problem that the opening and closing state of the isolating switch cannot be remotely identified in the prior art can be solved, and the effect of remotely and quickly judging the opening and closing state of the isolating switch can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings constituting part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0016] Figure 1 A schematic diagram of a three-column horizontal rotary isolating switch in a closed state provided by an embodiment of the present application is shown;

[0017] Figure 2 A schematic diagram of a three-column horizontal rotary isolating switch in an open state provided by an embodiment of the present application is shown;

[0018] Figure 3 A hardware structure block diagram of a mobile terminal for executing a method for detecting the opening and closing states of an isolating switch provided in an embodiment of the present application is shown;

[0019] Figure 4 A schematic flow chart of a method for detecting the opening and closing states of an isolating switch provided in an embodiment of the present application is shown;

[0020] Figure 5 A structural schematic diagram of a feature extraction and feature fusion module provided in an embodiment of the present application is shown;

[0021] Figure 6 A schematic diagram showing an enlarged current candidate region provided by an embodiment of the present application is shown;

[0022] Figure 7 A schematic diagram of a TBRNet network model structure provided by an embodiment of the present application is shown;

[0023] Figure 8 A coordinate schematic diagram of a target detection frame provided in an embodiment of the present application is shown;

[0024] Fig. 9 A schematic flow chart of a specific method for detecting the opening and closing states of an isolating switch provided in an embodiment of the present application is shown;

[0025] Fig.10 A structural block diagram of a device for detecting the opening and closing states of an isolating switch provided in an embodiment of the present application is shown.

[0026] The above drawings include the following reference numerals:

[0027] 102, processor; 104, memory; 106, transmission device; 108, input and output devices. DETAILED DESCRIPTION

[0028] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0029] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] The conductive part of the three-column horizontal rotary disconnector is fixed on the upper ends of the three post insulators. The post insulators on both sides are fixed, and the middle post insulator is rotating. The main conductive part consists of a conductive arm and static contacts on both sides. The conductive arm is installed on the middle post insulator, and the static contacts on both sides are installed on the insulators on both sides. The operating mechanism drives the middle post insulator to rotate to complete the opening and closing actions of the disconnector. When the disconnector is opened, the main shaft of the mechanism drives the middle rotating insulator to rotate 90° through the connecting rod transmission, so that the moving contact on the conductive arm is separated from the static contacts on both sides, forming a clearly visible isolation fracture. When the disconnector is to be closed, the middle rotating insulator rotates to make the moving contact on the conductive arm contact with the static contacts on the post insulators on both sides, forming a stable conductive path, thereby achieving the closing effect. Figure 1 As shown in the figure, the three-column horizontal rotary disconnector is in the closed state, where AB is the conductive arm of the disconnector, and C and D are the static contacts on both sides of the disconnector respectively; Figure 2 As shown, the conductive arm of the three-column horizontal rotary disconnector is rotated 90° and is in the open state.

[0032] As introduced in the background technology, the prior art is unable to remotely identify the open and closed states of the isolating switch. In order to solve the problem that the open and closed states of the isolating switch cannot be remotely identified, the embodiments of the present application provide a method, device, computer-readable storage medium and electronic device for detecting the open and closed states of the isolating switch.

[0033] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0034] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 3 1 is a hardware structure block diagram of a mobile terminal of a method for detecting the disconnection and closing states of an isolating switch according to an embodiment of the present invention. Figure 3 As shown, the mobile terminal may include one or more ( Figure 3Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 3 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 3 More or fewer components as shown, or with Figure 3 Different configurations shown.

[0035] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the detection method of the disconnecting and closing state of the disconnecting switch in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0036] In this embodiment, a method for detecting the open and closed states of an isolating switch running on a mobile terminal, a computer terminal or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0037] Figure 4 1 is a flow chart of a method for detecting the disconnection and closing states of an isolating switch according to an embodiment of the present application. Figure 4 As shown, the method comprises the following steps:

[0038] Step S201, acquiring disconnector image data, wherein the disconnector image data is image data including at least the disconnector, and the disconnector includes at least a conductive arm and a static contact;

[0039] Specifically, images of disconnectors are taken in different directions, different time periods, different scenes and different lighting conditions in the substation, and these images are subjected to image enhancement operations such as rotation, flipping, scaling and distortion to obtain more disconnector images. All disconnector image data constitute an disconnector image data set. In the following text, the disconnectors are all three-column horizontal rotary disconnectors, and the candidate areas described in the following text are all candidate areas of the three-column horizontal rotary disconnector images. In actual application, it can also be used for the detection of the opening and closing of other disconnectors.

[0040] Step S202, analyzing the isolating switch image data through a target detection network, generating a conductive arm detection frame and a static contact detection frame, extracting the coordinates of the target position of the conductive arm detection frame to obtain a first coordinate, and extracting the coordinates of the target position of the static contact detection frame to obtain a second coordinate, wherein the conductive arm detection frame at least includes an image of the conductive arm, and the static contact detection frame at least includes an image of the static contact, and the target detection network is obtained by training multiple sets of data, and each set of the multiple sets of data includes historical image data of the isolating switch and a conductive arm detection frame label and a static contact detection frame label;

[0041] Specifically, a target detection network is designed, which can be a TBRNet network, for generating a target detection frame. It includes a feature extraction and feature fusion module, which processes the isolating switch image through multiple layers of convolution, and fuses the features extracted by the multiple layers of convolution to enrich the fused feature information; then, a candidate frame extraction network suitable for the isolating switch image data set is used to generate a candidate area and extract the features of the candidate area; then, the global spatial features and local spatial features of the candidate area are combined to generate a target detection frame, which includes the above-mentioned conductive arm detection frame and static contact detection frame in this application. In actual application, it can also be used for the detection of other targets.

[0042] Step S203, calculate the distance between the first coordinate and the second coordinate, and when the distance is greater than a first preset threshold, determine that the disconnector is in an open state; when the distance is less than or equal to a second preset threshold, determine that the disconnector is in a closed state, wherein the first preset threshold is greater than the second preset threshold.

[0043] Specifically, after the detection frame of the conductive arm of the disconnector and the detection frames of the static contacts on both sides of the disconnector are generated by the trained TBRNet network model, the coordinates of the detection frame of the conductive arm of the disconnector and the coordinates of the detection frames of the static contacts on both sides of the disconnector are extracted first, and then the distance is calculated as the judgment basis for the subsequent disconnector status detection to determine whether the disconnector status is closed or open.

[0044] Through this embodiment, the isolating switch image data is obtained, and the isolating switch image data is image data that at least includes the isolating switch, and the isolating switch at least includes a conductive arm and a static contact; the isolating switch image data is analyzed by the target detection network, and a conductive arm detection frame and a static contact detection frame are generated, and the coordinates of the target position of the conductive arm detection frame are extracted to obtain a first coordinate, and the coordinates of the target position of the static contact detection frame are extracted to obtain a second coordinate, and the distance between the first coordinate and the second coordinate is calculated. When the distance is greater than the first preset threshold, the isolating switch is determined to be in an open state, and when the distance is less than or equal to the second preset threshold, the isolating switch is determined to be in a closed state. Compared with the prior art, which cannot remotely identify the opening and closing state of the isolating switch, the present application uses a marking frame to mark the conductive arm and the static contact of the isolating switch through the target detection network, and determines the opening and closing state of the isolating switch according to the distance of the coordinates of the marking frame, which can be remotely identified and judged, and the accuracy of identification and judgment is improved. Therefore, it can solve the problem that the opening and closing state of the isolating switch cannot be remotely identified in the prior art, and achieve the effect of remotely and quickly judging the opening and closing state of the isolating switch.

[0045] In the specific implementation process, before the above step S201 analyzes the isolating switch image data through the target detection network, the method further includes the following steps: obtaining the historical image data of the isolating switch; annotating the conductive arm of the isolating switch in the historical image data with a rotating frame to obtain a conductive arm detection frame, and annotating the vertex coordinates of the conductive arm detection frame to obtain the conductive arm detection frame label; annotating the static contact of the isolating switch with a horizontal frame to obtain a static contact detection frame, and annotating the vertex coordinates of the static contact detection frame to obtain the static contact detection frame label, wherein the rotating frame is obtained by rotating the horizontal frame; obtaining an initial target detection network, and training the initial target detection network with the historical image data, the conductive arm detection frame label and the static contact detection frame label to obtain the target detection network. The method trains the target detection network through the above steps, so that the network can be trained to recognize targets through target detection frames.

[0046] Specifically, during the annotation process, the image of the disconnector is annotated using annotation software such as roLabelImg. First, the conductive arm of the disconnector is annotated with a rotating frame, and the static contacts on both sides of the disconnector are annotated with a horizontal frame. After each image is annotated, it is saved and an XML file is automatically generated, which contains the coordinate information of the rotating frame and the horizontal frame, and each image corresponds to the generated XML file. In the XML file after the annotation is completed, the rotating frame and the horizontal frame are uniformly represented as an octet (x0, y0, x1, y1, x2, y2, x3, y3), where (x0, y0), (x1, y1), (x2, y2), and (x3, y3) are the coordinates of the four vertices of the frame respectively.

[0047] In some optional embodiments, the above-mentioned step S202 analyzes the isolating switch image data through the target detection network to generate a conductive arm detection frame and a static contact detection frame, which can be achieved by the following steps: Step S2021: extracting the appearance features, difference features and position features of the isolating switch through the target detection network, and performing feature fusion on the appearance features, difference features and position features to obtain a feature image after feature fusion, wherein the appearance features are features that characterize the appearance of the isolating switch, the difference features are features that characterize the differences between the isolating switches of different categories, and the position features are features that characterize the position changes of the isolating switch; Step S2022: obtaining the average size of the isolating switch in the isolating switch image data, and performing feature fusion on the target detection network. The size of the detection frame is adjusted to the average size, and the candidate area of ​​the feature image is generated through the detection frame, wherein the detection frame is used to detect the conductive arm and the static contact, and the candidate area is the area where the conductive arm and the static contact are located, which is preliminarily determined; step S2023: extract the features of the candidate area to obtain the candidate area features, extract the global space features and the local space features in the candidate area features, fuse the global space features and the local space features, and update the candidate area features, thereby generating the conductive arm detection frame and the static contact detection frame, wherein the candidate area features include all features in the candidate area, the global space features are features that characterize the conductive arm and the static contact, and the local space features are features that characterize the image background. The method generates a target detection frame through the above steps, so that the conductive arm and the static contact can be detected respectively through the target detection frame.

[0048] During the specific implementation process, a feature extraction and feature fusion module is designed in the target detection network to obtain richer feature information.

[0049] like Figure 5As shown in the figure, the feature extraction and feature fusion module consists of 9 convolutional layers, 9 ReLU layers and 2 maximum pooling layers. The nine convolutional layers are divided into five groups, namely Con1, Con2, Con3, Con4 and Con5. All convolutional layers use 3×3 convolution kernels to extract features, the convolution step is 1, and the ReLU activation function is used after each convolution layer. The two groups of convolution operations, Con1 and Con5, are followed by a maximum pooling layer with a pooling window of 2×2 and a pooling step of 2. At the same time, a residual connection structure is added to Con2 and Con4, Con3 and Con5 to solve the gradient disappearance or gradient explosion problem of this module.

[0050] After five sets of convolutions, the feature maps F2, F4, and F5 obtained by the three sets of convolution operations Con2, Con4, and Con5 are fused to obtain richer feature information. First, F2 and F4 are downsampled to the same size as F5 to obtain feature maps F2' and F4', and then feature fusion (concat) operation is performed to add the feature elements of the corresponding positions of the three feature maps F2', F4', and F5' and calculate the average value to obtain the fused feature map.

[0051] The two groups of convolution operations, Con1 and Con2, learned the basic information of the isolating switch, such as the outline, edge, and color. The convolution operation of Con3 learned the texture and material characteristics of the isolating switch. The convolution operation of Con4 learned the significant difference characteristics between the two-class problem of isolating switches. The convolution operation of Con5 learned the complete and identifiable key features of the isolating switch, and obtained the information of the position change of the isolating switch. Therefore, only the features after the convolution operation of Con2, Con4 and Con5 were selected for fusion.

[0052] Design a candidate box extraction network suitable for the disconnector image dataset to generate candidate regions. The average size and aspect ratio of the boxes annotated by the disconnector conductive arm and the static contacts on both sides in the statistical data preprocessing are calculated, and then the size and aspect ratio of the anchor box in the RPN network are changed to the size and aspect ratio of the anchor box suitable for the disconnector image dataset. After statistics, the size and aspect ratio of the anchor box suitable for the disconnector image dataset are approximately (32, 64, 128) and (1:1, 3:1, 4:1). Therefore, these sizes and aspect ratios are used as the size and aspect ratio of the anchor box in the RPN network to obtain a candidate box extraction network suitable for the disconnector image dataset. The size and aspect ratio of the anchor box are continuously adjusted and optimized during the network training process. Then, the network is used to generate candidate regions, and the candidate region features are extracted through candidate region pooling (RoI-Pooling). RPN (Region Proposal Network) is a neural network module for target detection. It is usually embedded in the entire target detection network as a subnetwork. Its main function is to generate candidate target regions. The global spatial features and local spatial features of the candidate region are extracted, and then the candidate region features are updated and new candidate region features are generated.

[0053] In some optional implementations, step S2023 extracts the global spatial features and local spatial features from the candidate region features, fuses the global spatial features and the local spatial features, and updates the candidate region features, which can be achieved by the following steps: step S2024: divide the candidate region features into current candidate region features and other candidate region features; step S2025: extract the global spatial features through the current candidate region features, and extract the local spatial features through other candidate region features; step S2026: fuse the current candidate region features with the global spatial features to obtain a first fused feature, fuse the current candidate region features with the local spatial features to obtain a second fused feature, fuse the first fused feature with the second fused feature to obtain the updated candidate region features. The method updates the candidate region features through the above steps, so that the features extracted by the network can be made more accurate through iterative updates.

[0054] In the specific implementation process, the extracted candidate region features are divided into current candidate region features and other candidate region features, and the global spatial features f are extracted through the current candidate region features G , and extract local spatial features f through other candidate region features L , then f G and f L Through the attention mechanism and the current candidate region feature f vThe two updated candidate region features are fused and then the two features are fused to generate a new candidate region feature F.

[0055] In some optional implementations, the above step S2022 extracts the global space feature through the feature of the current candidate region, which can be implemented by the following steps: step S2027: obtaining the expansion coefficient, and calculating the extended vertex coordinates according to the vertex coordinates of the current candidate region and the expansion coefficient; step S2028: obtaining the height and width of the isolating switch image, and updating the extended vertex coordinates according to the height and the width to obtain the extended candidate region; step S2029: extracting the feature of the extended candidate region to obtain the global space feature. The method extracts the global space feature through the above steps, so that the features corresponding to the target object in the image of the entire isolating switch can be extracted.

[0056] In the specific implementation process, the global spatial features can be extracted by expanding the current candidate area so that the area contains the conductive arm of the disconnector and the static contacts on both sides that need to be detected as much as possible. Figure 6 As shown, let the enlarged area be R, and the enlargement coefficient be θ, which means that the height and width of R are enlarged to θ times of the original current candidate area. Among them, (x0, y0), (x1, y1), (x2, y2), (x3, y3) are the coordinates of the four vertices of the current candidate area respectively. They are the coordinates of the four vertices of the enlarged region R.

[0057] The coordinate calculation formula of region R is as follows:

[0058]

[0059]

[0060] Where θ is the expansion coefficient, h and w are the height and width of the current candidate area:

[0061] h=y2-y1

[0062] w=x1-x0

[0063] After obtaining the coordinates of region R, in order to prevent the coordinates from overflowing the boundary of the isolating switch image, the coordinates of region R need to be updated. Assuming the height and width of the isolating switch image are H and W, the update formula is as follows:

[0064]

[0065] Finally, the features of region R are extracted through the RoI-Pooling operation, thereby realizing the extraction of global spatial features.

[0066] Extract local spatial features through the feature information of other candidate regions E. Different candidate regions e in E i The contribution to extracting local spatial features is different. The weight parameter a is set i Represents different candidate regions e i The contribution of i is the candidate region e i The characteristics of a i The calculation formula is as follows:

[0067] a i =softmax(tanh(W e ·f i )·relu(w f ·R iv ))

[0068] Among them, softmax is the normalization function, W e and W f is the learnable weight matrix, R iv Represents the candidate region e i The spatial position relationship with the current candidate region v is a scalar:

[0069]

[0070] Where (x v ,y v ) is the vertex coordinate of the current candidate region v, (x i ,y i ) is the candidate region e i The vertex coordinates, h v ,w v ,h i ,w i are v and e respectively i The height and width of the

[0071] Finally, calculate the local spatial feature f L The calculation formula is as follows:

[0072]

[0073] Extract the global space feature f G and the local spatial feature f L After that, G and f L The features and f are obtained through the attention mechanism module respectively. L , where f G ' and f L 'Characteristic ratio f G and f L Features can pay more attention to the detailed information of the detected target.

[0074] Then f G 'With the current regional features f v Fusion to obtain updated regional features f L ' and f v Fusion to obtain updated regional features and The calculation formula is as follows:

[0075]

[0076] Finally, the updated features are used to generate new candidate region features F. The learning weight parameter β is added in the generation process, and β is adjusted by dynamically adjusting β (between 0 and 1) during the network training process. and The weight of F is calculated as follows:

[0077]

[0078] Construct the TBRNet network model structure and generate the target detection box.

[0079] The TBRNet network model structure is as follows Figure 7 As shown in the figure, the three-column horizontal rotary disconnector image is input, and the feature map is obtained through the feature extraction and feature fusion modules. The candidate region features are obtained through the candidate frame extraction network and RoI-Pooling operation. The global spatial features and local spatial features are extracted from the candidate region features, and then the candidate region features are updated and new candidate region features are generated. Finally, the target detection frame is generated by regression.

[0080] The disconnector image dataset is divided into training dataset, validation dataset and test dataset in the ratio of 8:1:1, and then the TBRNet network is trained. First, the network weight parameters and network hyperparameters are randomly initialized and set, such as the learning rate size, regularization parameters, etc., and then the stochastic gradient descent optimization algorithm is selected to update the network model parameters. During the training process, the validation dataset is periodically used to evaluate the performance of the network model, so as to adjust the network model structure and hyperparameters in time and enable the network to converge quickly. Then a trained TBRNet network model that can generate a target detection frame is obtained, and the disconnector image is input into the network model, and then the target detection frame is generated.

[0081] In some optional embodiments, there are two static contacts, and the two static contacts are respectively located on both sides of the conductive arm. The above step S202 extracts the coordinates of the target position of the conductive arm detection frame to obtain the first coordinate, and extracts the coordinates of the target position of the static contact detection frame to obtain the second coordinate. This can be achieved by the following steps: extracting the coordinates of the lower right corner position of the conductive arm detection frame to obtain the first coordinate; extracting the coordinates of the lower right corner position of the two static contact detection frames to obtain the second coordinate corresponding to each static contact detection frame diagram. The method determines the coordinates of the detection frame through the above steps, which is convenient for subsequent calculation of the distance between the target detection objects and judgment of the opening and closing state.

[0082] In the specific implementation process, Figure 8 As shown in the figure, the target position selects the lower right corner of the target detection frame, and extracts the coordinate A (x a ,y a ) and the coordinates B(x b ,y b ) and C(x c ,y c ). In actual application, the position of the detection frame can be determined according to the object that actually needs to be detected.

[0083] In some optional embodiments, the above step S203 calculates the distance between the first coordinate and the second coordinate, and when the distance is greater than the first preset threshold, determines that the disconnector is in the open state, and when the distance is less than or equal to the second preset threshold, determines that the disconnector is in the closed state, which can be achieved by the following steps: respectively calculating the distance between the first coordinate and each of the second coordinates to obtain the first distance and the second distance; determining the minimum distance between the first distance and the second distance, and when the minimum distance is greater than the first preset threshold, determines that the disconnector is in the open state, and when the minimum distance is less than or equal to the second preset threshold, determines that the disconnector is in the closed state. The method determines the open and closed state of the disconnector through the above steps, so that the state of the disconnector can be accurately judged.

[0084] Specifically, the distances d1 and d2 between AB and AC are calculated, and the minimum distance d between them is min As the basis for subsequent disconnector status detection, the calculation formula is as follows:

[0085]

[0086] d min =min(d1,d2)

[0087] Finally, according to the results of multiple tests, d min The disconnector state in the interval [12,40] is identified as "closed state", and d min The isolating switch state in the interval [65, 150] is identified as the “open state.” The first preset threshold and the second preset threshold can be adjusted according to actual conditions.

[0088] In order to enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the method for detecting the opening and closing states of the disconnecting switch of the present application will be described in detail below in combination with specific embodiments.

[0089] This embodiment relates to a specific method for detecting the opening and closing states of an isolating switch, such as Fig. 9 As shown, the following steps are included:

[0090] Step 1: Construct a three-column horizontal rotary disconnector image dataset;

[0091] Step 2: Data preprocessing;

[0092] Step 3: Design a TBRNet network to generate the target detection frame. First, perform multi-layer convolution on the disconnector image and fuse the features extracted by the multi-layer convolution. Then, generate the candidate region through the candidate frame extraction network and extract the candidate region features. Then, combine the global spatial features and local spatial features of the candidate region to generate the final target detection frame.

[0093] Step 4: Divide the constructed data set and train the TBRNet network, and finally obtain a network model for generating target detection boxes;

[0094] Step 5: Identify the open and closed status of the disconnector.

[0095] The embodiment of the present application also provides a detection device for the disconnection and closing state of the disconnector. It should be noted that the detection device for the disconnection and closing state of the disconnector in the embodiment of the present application can be used to execute the detection method for the disconnection and closing state of the disconnector provided in the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0096] The following is an introduction to the detection device for the opening and closing status of the isolating switch provided in an embodiment of the present application.

[0097] Fig.10 Schematic diagram of a detection device for the disconnecting and closing state of an isolating switch according to an embodiment of the present application. Fig.10 As shown, the device comprises:

[0098] A first acquisition unit 10 is used to acquire image data of the isolating switch, wherein the image data of the isolating switch is image data including at least the isolating switch, and the isolating switch includes at least a conductive arm and a static contact;

[0099] Specifically, images of disconnectors are taken in different directions, different time periods, different scenes and different lighting conditions in the substation, and these images are subjected to image enhancement operations such as rotation, flipping, scaling and distortion to obtain more disconnector images. All disconnector image data constitute an disconnector image data set. In the following text, the disconnectors are all three-column horizontal rotary disconnectors, and the candidate areas described in the following text are all candidate areas of the three-column horizontal rotary disconnector images. In actual application, it can also be used for the detection of the opening and closing of other disconnectors.

[0100] an extraction unit 20, configured to analyze the isolating switch image data through a target detection network, generate a conductive arm detection frame and a static contact detection frame, extract the coordinates of the target position of the conductive arm detection frame to obtain a first coordinate, and extract the coordinates of the target position of the static contact detection frame to obtain a second coordinate, wherein the conductive arm detection frame at least includes an image of the conductive arm, and the static contact detection frame at least includes an image of the static contact, and the target detection network is obtained by training multiple sets of data, and each set of the multiple sets of data includes historical image data of the isolating switch and a conductive arm detection frame label and a static contact detection frame label;

[0101] Specifically, a target detection network is designed, which can be a TBRNet network, for generating a target detection frame. It includes a feature extraction and feature fusion module, which processes the isolating switch image through multiple layers of convolution, and fuses the features extracted by the multiple layers of convolution to enrich the fused feature information; then, a candidate frame extraction network suitable for the isolating switch image data set is used to generate a candidate area and extract the features of the candidate area; then, the global spatial features and local spatial features of the candidate area are combined to generate a target detection frame, which includes the above-mentioned conductive arm detection frame and static contact detection frame in this application. In actual application, it can also be used for the detection of other targets.

[0102] The determination unit 30 is used to calculate the distance between the first coordinate and the second coordinate, and determine that the disconnector is in an open state when the distance is greater than a first preset threshold value, and determine that the disconnector is in a closed state when the distance is less than or equal to a second preset threshold value, wherein the first preset threshold value is greater than the second preset threshold value.

[0103] Specifically, after the detection frame of the conductive arm of the disconnector and the detection frames of the static contacts on both sides of the disconnector are generated by the trained TBRNet network model, the coordinates of the detection frame of the conductive arm of the disconnector and the coordinates of the detection frames of the static contacts on both sides of the disconnector are extracted first, and then the distance is calculated as the judgment basis for the subsequent disconnector status detection to determine whether the disconnector status is closed or open.

[0104] Through this embodiment, the isolating switch image data is obtained, and the isolating switch image data is image data that at least includes the isolating switch, and the isolating switch at least includes a conductive arm and a static contact; the isolating switch image data is analyzed by the target detection network, and a conductive arm detection frame and a static contact detection frame are generated, and the coordinates of the target position of the conductive arm detection frame are extracted to obtain a first coordinate, and the coordinates of the target position of the static contact detection frame are extracted to obtain a second coordinate, and the distance between the first coordinate and the second coordinate is calculated. When the distance is greater than the first preset threshold, the isolating switch is determined to be in an open state, and when the distance is less than or equal to the second preset threshold, the isolating switch is determined to be in a closed state. Compared with the prior art, which cannot remotely identify the opening and closing state of the isolating switch, the present application uses a marking frame to mark the conductive arm and the static contact of the isolating switch through the target detection network, and determines the opening and closing state of the isolating switch according to the distance of the coordinates of the marking frame, which can be remotely identified and judged, and the accuracy of identification and judgment is improved. Therefore, it can solve the problem that the opening and closing state of the isolating switch cannot be remotely identified in the prior art, and achieve the effect of remotely and quickly judging the opening and closing state of the isolating switch.

[0105] In the specific implementation process, before the above step S201 analyzes the isolating switch image data through the target detection network, the device also includes a second acquisition unit, a first labeling unit, a second labeling unit and a training unit, the second acquisition unit is used to acquire the historical image data of the isolating switch; the first labeling unit is used to label the conductive arm of the isolating switch in the historical image data with a rotating frame to obtain a conductive arm detection frame, and to label the vertex coordinates of the conductive arm detection frame to obtain the conductive arm detection frame label; the second labeling unit is used to label the static contact of the isolating switch with a horizontal frame to obtain a static contact detection frame, and to label the vertex coordinates of the static contact detection frame to obtain the static contact detection frame label, wherein the rotating frame is obtained by rotating the horizontal frame; the training unit is used to acquire an initial target detection network, and to train the initial target detection network with the historical image data, the conductive arm detection frame label and the static contact detection frame label to obtain the target detection network. The device trains the target detection network through the above steps, so that the network can be trained to recognize targets through target detection frames.

[0106] Specifically, during the annotation process, the image of the disconnector is annotated using annotation software such as roLabelImg. First, the conductive arm of the disconnector is annotated with a rotating frame, and the static contacts on both sides of the disconnector are annotated with a horizontal frame. After each image is annotated, it is saved and an XML file is automatically generated, which contains the coordinate information of the rotating frame and the horizontal frame, and each image corresponds to the generated XML file. In the XML file after the annotation is completed, the rotating frame and the horizontal frame are uniformly represented as an octet (x0, y0, x1, y1, x2, y2, x3, y3), where (x0, y0), (x1, y1), (x2, y2), and (x3, y3) are the coordinates of the four vertices of the frame respectively.

[0107] In some optional embodiments, the extraction unit includes a fusion module, a first generation module and a second generation module, the fusion module is used to extract the appearance features, difference features and position features of the isolating switch through the target detection network, and perform feature fusion on the appearance features, difference features and position features to obtain a feature image after feature fusion, wherein the appearance features are features that characterize the appearance of the isolating switch, the difference features are features that characterize the differences between the isolating switches of different categories, and the position features are features that characterize the position changes of the isolating switch; the first generation module is used to obtain the average size of the isolating switches in the isolating switch image data, and adjust the size of the detection frame of the target detection network to the average size, The candidate area of ​​the feature image is generated through the detection frame, wherein the detection frame is used to detect the conductive arm and the static contact, and the candidate area is the area where the conductive arm and the static contact are located, which is preliminarily determined; the second generation module is used to extract the features of the candidate area, obtain the candidate area features, extract the global space features and local space features in the candidate area features, fuse the global space features and the local space features, and update the candidate area features, thereby generating the conductive arm detection frame and the static contact detection frame, wherein the candidate area features include all features in the candidate area, the global space features are features that characterize the conductive arm and the static contact, and the local space features are features that characterize the image background. The device generates a target detection frame through the above steps, so that the conductive arm and the static contact can be detected separately through the target detection frame.

[0108] During the specific implementation process, a feature extraction and feature fusion module is designed in the target detection network to obtain richer feature information.

[0109] like Figure 5As shown in the figure, the feature extraction and feature fusion module consists of 9 convolutional layers, 9 ReLU layers and 2 maximum pooling layers. The nine convolutional layers are divided into five groups, namely Con1, Con2, Con3, Con4 and Con5. All convolutional layers use 3×3 convolution kernels to extract features, the convolution step is 1, and the ReLU activation function is used after each convolution layer. The two groups of convolution operations, Con1 and Con5, are followed by a maximum pooling layer with a pooling window of 2×2 and a pooling step of 2. At the same time, a residual connection structure is added to Con2 and Con4, Con3 and Con5 to solve the gradient disappearance or gradient explosion problem of this module.

[0110] After five sets of convolutions, the feature maps F2, F4, and F5 obtained by the three sets of convolution operations Con2, Con4, and Con5 are fused to obtain richer feature information. First, F2 and F4 are downsampled to the same size as F5 to obtain feature maps F2' and F4', and then feature fusion (concat) operation is performed to add the feature elements of the corresponding positions of the three feature maps F2', F4', and F5' and calculate the average value to obtain the fused feature map.

[0111] The two groups of convolution operations, Con1 and Con2, learned the basic information of the isolating switch, such as the outline, edge, and color. The convolution operation of Con3 learned the texture and material characteristics of the isolating switch. The convolution operation of Con4 learned the significant difference characteristics between the two-class problem of isolating switches. The convolution operation of Con5 learned the complete and identifiable key features of the isolating switch, and obtained the information of the position change of the isolating switch. Therefore, only the features after the convolution operation of Con2, Con4 and Con5 were selected for fusion.

[0112] Design a candidate box extraction network suitable for the disconnector image dataset to generate candidate regions. The average size and aspect ratio of the boxes annotated by the disconnector conductive arm and the static contacts on both sides in the statistical data preprocessing are calculated, and then the size and aspect ratio of the anchor box in the RPN network are changed to the size and aspect ratio of the anchor box suitable for the disconnector image dataset. After statistics, the size and aspect ratio of the anchor box suitable for the disconnector image dataset are approximately (32, 64, 128) and (1:1, 3:1, 4:1). Therefore, these sizes and aspect ratios are used as the size and aspect ratio of the anchor box in the RPN network to obtain a candidate box extraction network suitable for the disconnector image dataset. The size and aspect ratio of the anchor box are continuously adjusted and optimized during the network training process. Then, the network is used to generate candidate regions, and the candidate region features are extracted through candidate region pooling (RoI-Pooling). RPN (Region Proposal Network) is a neural network module for target detection. It is usually embedded in the entire target detection network as a subnetwork. Its main function is to generate candidate target regions. The global spatial features and local spatial features of the candidate region are extracted, and then the candidate region features are updated and new candidate region features are generated.

[0113] In some optional implementations, the second generation module includes a division submodule, an extraction submodule, and a fusion submodule. The division submodule is used to divide the candidate region features into current candidate region features and other candidate region features; the extraction submodule is used to extract the global spatial features through the current candidate region features, and extract the local spatial features through other candidate region features; the fusion submodule is used to fuse the current candidate region features with the global spatial features to obtain a first fused feature, fuse the current candidate region features with the local spatial features to obtain a second fused feature, and fuse the first fused feature with the second fused feature to obtain the updated candidate region features. The device updates the candidate region features through the above steps, so that the features extracted by the network can be made more accurate through iterative updates.

[0114] In the specific implementation process, the extracted candidate region features are divided into current candidate region features and other candidate region features, and the global spatial features f are extracted through the current candidate region features G , and extract local spatial features f through other candidate region features L , then f G and f L Through the attention mechanism and the current candidate region feature f v The two updated candidate region features are fused and then the two features are fused to generate a new candidate region feature F.

[0115] In some optional embodiments, the extraction submodule includes a calculation submodule, an expansion submodule and an extraction submodule, wherein the calculation submodule is used to obtain the expansion coefficient, and calculate the expansion vertex coordinates according to the vertex coordinates of the current candidate area and the expansion coefficient; the expansion submodule is used to obtain the height and width of the disconnector image, and update the expansion vertex coordinates according to the height and the width to obtain the expansion candidate area; the extraction submodule is used to extract the features of the expansion candidate area to obtain the global space features. The device extracts the global space features through the above steps, so that the features corresponding to the target object in the image of the entire disconnector can be extracted.

[0116] In the specific implementation process, the global spatial features can be extracted by expanding the current candidate area so that the area contains the conductive arm of the disconnector and the static contacts on both sides that need to be detected as much as possible. Figure 6 As shown, let the enlarged area be R, and the enlargement coefficient be θ, which means that the height and width of R are enlarged to θ times of the original current candidate area. Among them, (x0, y0), (x1, y1), (x2, y2), (x3, y3) are the coordinates of the four vertices of the current candidate area respectively. They are the coordinates of the four vertices of the enlarged region R.

[0117] The coordinate calculation formula of region R is as follows:

[0118]

[0119] Where θ is the expansion coefficient, h and w are the height and width of the current candidate area:

[0120] h=y2-y1

[0121] w=x1-x0

[0122] After obtaining the coordinates of region R, in order to prevent the coordinates from overflowing the boundary of the isolating switch image, the coordinates of region R need to be updated. Assuming the height and width of the isolating switch image are H and W, the update formula is as follows:

[0123]

[0124] Finally, the features of region R are extracted through the RoI-Pooling operation, thereby realizing the extraction of global spatial features.

[0125] Extract local spatial features through the feature information of other candidate regions E. Different candidate regions e in E i The contribution to extracting local spatial features is different. The weight parameter a is set i Represents different candidate regions e i The contribution of i is the candidate region e iThe characteristics of a i The calculation formula is as follows:

[0126] a i =softmax(tanh(W e ·f i )·relu(w f ·R iv ))

[0127] Among them, softmax is the normalization function, W e and W f is the learnable weight matrix, R iv Represents the candidate region e i The spatial position relationship with the current candidate region v is a scalar:

[0128]

[0129] Where (x v ,y v ) is the vertex coordinate of the current candidate region v, (x i ,y i ) is the candidate region e i The vertex coordinates, h v ,w v ,h i ,w i are v and e respectively i The height and width of the

[0130] Finally, calculate the local spatial feature f L The calculation formula is as follows:

[0131]

[0132] Extract the global space feature f G and the local spatial feature f L After that, G and f L The features and f are obtained through the attention mechanism module respectively. L , where f G ' and f L 'Characteristic ratio f G and f L Features can pay more attention to the detailed information of the detected target.

[0133] Then f G 'With the current regional features f v Fusion to obtain updated regional features f L ' and f v Fusion to obtain updated regional features and The calculation formula is as follows:

[0134]

[0135] Finally, the updated features are used to generate new candidate region features F. The learning weight parameter β is added in the generation process, and β is adjusted by dynamically adjusting β (between 0 and 1) during the network training process. and The weight of F is calculated as follows:

[0136]

[0137] Construct the TBRNet network model structure and generate the target detection box.

[0138] The TBRNet network model structure is as follows Figure 7 As shown in the figure, the three-column horizontal rotary disconnector image is input, and the feature map is obtained through the feature extraction and feature fusion modules. The candidate region features are obtained through the candidate frame extraction network and RoI-Pooling operation. The global spatial features and local spatial features are extracted from the candidate region features, and then the candidate region features are updated and new candidate region features are generated. Finally, the target detection frame is generated by regression.

[0139] The disconnector image dataset is divided into training dataset, validation dataset and test dataset in the ratio of 8:1:1, and then the TBRNet network is trained. First, the network weight parameters and network hyperparameters are randomly initialized and set, such as the learning rate size, regularization parameters, etc., and then the stochastic gradient descent optimization algorithm is selected to update the network model parameters. During the training process, the validation dataset is periodically used to evaluate the performance of the network model, so as to adjust the network model structure and hyperparameters in time and enable the network to converge quickly. Then a trained TBRNet network model that can generate a target detection frame is obtained, and the disconnector image is input into the network model, and then the target detection frame is generated.

[0140] In some optional embodiments, there are two static contacts, and the two static contacts are respectively located on both sides of the conductive arm. The extraction unit includes a first extraction module and a second extraction module. The first extraction module is used to extract the coordinates of the lower right corner position of the conductive arm detection frame to obtain the first coordinate; the second extraction module is used to extract the coordinates of the lower right corner position of the two static contact detection frames to obtain the second coordinate corresponding to each static contact detection frame diagram. The device determines the coordinates of the detection frame through the above steps, which is convenient for subsequent calculation of the distance between the target detection objects and judgment of the opening and closing state.

[0141] In the specific implementation process, Figure 8As shown in the figure, the target position selects the lower right corner of the target detection frame, and extracts the coordinate A (x a ,y a ) and the coordinates B(x b ,y b ) and C(x c ,y c ). In actual application, the position of the detection frame can be determined according to the object that actually needs to be detected.

[0142] In some optional embodiments, the determination unit includes a calculation module and a determination module, the calculation module is used to calculate the distance between the first coordinate and each of the second coordinates respectively to obtain the first distance and the second distance; the determination module is used to determine the minimum distance between the first distance and the second distance, and when the minimum distance is greater than the first preset threshold, the disconnector is determined to be in the open state, and when the minimum distance is less than or equal to the second preset threshold, the disconnector is determined to be in the closed state. The device determines the open and closed state of the disconnector through the above steps, so that the state of the disconnector can be accurately judged.

[0143] Specifically, the distances d1 and d2 between AB and AC are calculated, and the minimum distance d between them is min As the basis for subsequent disconnector status detection, the calculation formula is as follows:

[0144]

[0145] d min =min(d1,d2)

[0146] Finally, according to the results of multiple tests, d min The disconnector state in the interval [12,40] is identified as "closed state", and d min The isolating switch state in the interval [65, 150] is identified as the “open state.” The first preset threshold and the second preset threshold can be adjusted according to actual conditions.

[0147] The detection device for the disconnecting and closing state of the isolating switch includes a processor and a memory, and the first acquisition unit, the extraction unit and the determination unit are all stored in the memory as program units, and the processor executes the program units stored in the memory to realize the corresponding functions. The above modules are all located in the same processor; or, the above modules are located in different processors in any combination.

[0148] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the opening and closing states of the disconnector can be remotely identified by adjusting kernel parameters.

[0149] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0150] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored program, wherein when the program is run, the device where the computer-readable storage medium is located is controlled to execute the method for detecting the opening and closing states of the isolating switch.

[0151] An embodiment of the present invention provides an electronic device, including a processor, a memory, and a program stored in the memory and executable on the processor, and the above method is implemented when the processor executes the program.

[0152] The devices in this article can be servers, PCs, PADs, mobile phones, etc.

[0153] The present application also provides a computer program product, including a computer program, which implements the methods described in the various embodiments of the present application when executed by a processor.

[0154] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0155] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0156] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0157] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0158] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0159] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0160] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0161] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0162] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0163] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0164] 1) In the detection method of the disconnector opening and closing state of the present application, the disconnector image data is obtained, and the disconnector image data is image data including at least the disconnector, and the disconnector includes at least the conductive arm and the static contact; the disconnector image data is analyzed by the target detection network, and the conductive arm detection frame and the static contact detection frame are generated, and the coordinates of the target position of the conductive arm detection frame are extracted to obtain the first coordinate, and the coordinates of the target position of the static contact detection frame are extracted to obtain the second coordinate, and the distance between the first coordinate and the second coordinate is calculated. When the distance is greater than the first preset threshold, the disconnector is determined to be in the opening state, and when the distance is less than or equal to the second preset threshold, the disconnector is determined to be in the closing state. Compared with the prior art, which cannot remotely identify the opening and closing state of the disconnector, the present application uses a marking frame to mark the conductive arm and the static contact of the disconnector through the target detection network, and determines the opening and closing state of the disconnector according to the distance of the coordinates of the marking frame, which can be remotely identified and judged, and improves the accuracy of identification and judgment. Therefore, the problem in the prior art that the opening and closing states of the isolating switch cannot be remotely identified can be solved, and the effect of remotely and quickly judging the opening and closing states of the isolating switch can be achieved.

[0165] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for detecting the opening and closing state of an isolating switch, characterized in that: include: Acquire isolating switch image data, wherein the isolating switch image data is image data including at least the isolating switch, and the isolating switch includes at least a conductive arm and a static contact; The image data of the isolating switch is analyzed by a target detection network to generate a conductive arm detection frame and a static contact detection frame, and the coordinates of the target position of the conductive arm detection frame are extracted to obtain a first coordinate, and the coordinates of the target position of the static contact detection frame are extracted to obtain a second coordinate, wherein the conductive arm detection frame at least includes an image of the conductive arm, and the static contact detection frame at least includes an image of the static contact, and the target detection network is obtained by training multiple sets of data, and each set of the multiple sets of data includes historical image data of the isolating switch and a conductive arm detection frame label and a static contact detection frame label; Calculate the distance between the first coordinate and the second coordinate, and when the distance is greater than a first preset threshold, determine that the disconnector is in an open state; when the distance is less than or equal to a second preset threshold, determine that the disconnector is in a closed state, wherein the first preset threshold is greater than the second preset threshold.

2. The method for detecting the opening and closing state of the isolating switch according to claim 1 is characterized in that: Before analyzing the disconnector image data by the target detection network, the method further includes: Acquiring the historical image data of the disconnector; The conductive arm of the disconnector in the historical image data is marked with a rotating frame to obtain a conductive arm detection frame, and the vertex coordinates of the conductive arm detection frame are marked to obtain a conductive arm detection frame label; The stationary contact of the disconnector is marked with a horizontal frame to obtain a stationary contact detection frame, and the vertex coordinates of the stationary contact detection frame are marked to obtain a stationary contact detection frame label, wherein the rotation frame is obtained by rotating the horizontal frame; An initial target detection network is obtained, and the historical image data, the conductive arm detection frame label, and the static contact detection frame label are used to train the initial target detection network to obtain the target detection network.

3. The method for detecting the opening and closing state of the isolating switch according to claim 1 is characterized in that: The disconnector image data is analyzed by a target detection network to generate a conductive arm detection frame and a static contact detection frame, including: The appearance feature, difference feature and position feature of the disconnector are extracted by the target detection network, and the appearance feature, difference feature and position feature are feature fused to obtain a feature image after feature fusion, wherein the appearance feature is a feature characterizing the appearance of the disconnector, the difference feature is a feature characterizing the difference between the disconnectors of different categories, and the position feature is a feature characterizing the position change of the disconnector; Acquire an average size of the isolating switch in the isolating switch image data, adjust the size of the detection frame of the target detection network to the average size, and generate a candidate area of ​​the feature image through the detection frame, wherein the detection frame is used to detect the conductive arm and the stationary contact, and the candidate area is a preliminarily determined area where the conductive arm and the stationary contact are located; Extract the features of the candidate area to obtain the candidate area features, extract the global space features and the local space features in the candidate area features, fuse the global space features and the local space features, and update the candidate area features, so as to generate the conductive arm detection frame and the static contact detection frame, wherein the candidate area features include all features in the candidate area, the global space features are features that characterize the conductive arm and the static contact, and the local space features are features that characterize the image background.

4. The method for detecting the opening and closing state of the isolating switch according to claim 3 is characterized in that: Extracting global spatial features and local spatial features from the candidate region features, fusing the global spatial features with the local spatial features, and updating the candidate region features, including: Dividing the candidate region features into current candidate region features and other candidate region features; Extracting the global spatial feature through the current candidate region feature, and extracting the local spatial feature through other candidate region features; The current candidate region feature is fused with the global space feature to obtain a first fused feature, the current candidate region feature is fused with the local space feature to obtain a second fused feature, and the first fused feature and the second fused feature are fused to obtain the updated candidate region feature.

5. The method for detecting the opening and closing state of the isolating switch according to claim 4 is characterized in that: Extracting the global spatial feature through the current candidate region feature includes: Acquire an expansion coefficient, and calculate extended vertex coordinates according to vertex coordinates of the current candidate area and the expansion coefficient, wherein the expansion coefficient is used to expand the current candidate area; Acquire the height and width of the disconnector image, and update the extended vertex coordinates according to the height and the width to obtain an extended candidate area; The features of the extended candidate region are extracted to obtain the global spatial features.

6. The method for detecting the opening and closing state of the isolating switch according to claim 1 is characterized in that: There are two stationary contacts, which are respectively located on both sides of the conductive arm. The coordinates of the target position of the conductive arm detection frame are extracted to obtain the first coordinates. The coordinates of the target position of the stationary contact detection frame are extracted to obtain the second coordinates, including: Extracting the position coordinates of the lower right corner of the conductive arm detection frame to obtain the first coordinates; The lower right corner position coordinates of the two stationary contact detection frames are extracted to obtain the second coordinates corresponding to each stationary contact detection frame diagram.

7. The method for detecting the opening and closing state of the isolating switch according to claim 6 is characterized in that: Calculating the distance between the first coordinate and the second coordinate, and determining that the disconnector is in an open state when the distance is greater than a first preset threshold, and determining that the disconnector is in a closed state when the distance is less than or equal to a second preset threshold, includes: respectively calculating the distance between the first coordinate and each of the second coordinates to obtain a first distance and a second distance; Determine the minimum distance between the first distance and the second distance, and when the minimum distance is greater than the first preset threshold, determine that the disconnector is in the open state; when the minimum distance is less than or equal to the second preset threshold, determine that the disconnector is in the closed state.

8. A device for detecting the opening and closing state of an isolating switch, characterized in that: include: A first acquisition unit is used to acquire the isolating switch image data, wherein the isolating switch image data is image data including at least the isolating switch, and the isolating switch includes at least a conductive arm and a static contact; an extraction unit, configured to analyze the isolating switch image data through a target detection network, generate a conductive arm detection frame and a static contact detection frame, extract the coordinates of a target position of the conductive arm detection frame to obtain a first coordinate, and extract the coordinates of the target position of the static contact detection frame to obtain a second coordinate, wherein the conductive arm detection frame at least includes an image of the conductive arm, and the static contact detection frame at least includes an image of the static contact, and the target detection network is obtained by training multiple sets of data, and each set of the multiple sets of data includes historical image data of the isolating switch and a conductive arm detection frame label and a static contact detection frame label; A determination unit is used to calculate the distance between the first coordinate and the second coordinate, and when the distance is greater than a first preset threshold, determine that the disconnector is in an open state; when the distance is less than or equal to a second preset threshold, determine that the disconnector is in a closed state, wherein the first preset threshold is greater than the second preset threshold.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for detecting the opening and closing states of the disconnecting switch according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing the detection method of the disconnecting switch opening and closing state as described in any one of claims 1 to 7.

Images