Knife switch open / close state identification method, device, equipment and storage medium

Abstract

The present invention relates to the technical field of knife switch state detection, and discloses a knife switch opening and closing state identification method, device, equipment and storage medium. The method comprises: extracting point cloud feature parameters based on the detection signal emitted by a laser radar and the reflected echo signal and generating a point cloud image of the target area, extracting the structural edge point cloud data of the target knife switch in the point cloud image, constructing a three-dimensional image of the target knife switch, determining the position of each contact and each knife arm of the target knife switch in the three-dimensional image, and the slope of each knife arm; calculating the distance between each contact based on the position; and determining the current working state of the target knife switch based on the position, slope and distance. The present invention uses a laser radar to obtain the position information of each contact and each knife arm in the target knife switch, the slope of each knife arm and the distance between each contact, and determines the current working state of the target knife switch based on the position information, slope and distance, so as to achieve accurate identification of the opening and closing state of the knife switch under various environments and pixels.

Description

Technical Field

[0001] The present invention relates to the technical field of knife switch state detection, and in particular to a knife switch open / close state identification method, device, equipment and storage medium. Background Art

[0002] Existing technologies for detecting knife switch status primarily rely on video recognition and sensors. However, video recognition technology has high requirements for illumination and image pixel density during image capture. Recognition performance is poor at night, in rainy, foggy, or cloudy weather, and is prone to errors. Furthermore, video is a two-dimensional image and cannot accurately reflect the distance between the knife switch's conductive blade and static contact, making it difficult to accurately determine whether the switch is closed properly. Sensors are typically mounted on the knife switch structure, close to high-voltage live equipment. Installation and maintenance require power outages, which can be very complex and time-consuming. Summary of the Invention

[0003] The main purpose of the present invention is to solve the problem that the existing technology has high requirements on the environment and pixels when detecting the knife switch status, and cannot accurately identify the knife switch status when the environment is harsh or the pixels are low.

[0004] A first aspect of the present invention provides a method for identifying the open / closed state of a knife switch, the method comprising: using a laser radar to transmit a detection signal to a target area and receiving a reflected echo signal; comparing and analyzing the echo signal and the detection signal, extracting point cloud feature parameters related to the target area, and generating a point cloud image of the target area based on the point cloud feature parameters; using a pre-trained convolutional neural network to extract structural edge point cloud data of the target knife switch in the point cloud image, and constructing a three-dimensional image of the target knife switch; using a straight line segment detection method to determine the position of each contact and each knife arm of the target knife switch in the three-dimensional image, as well as the slope of each knife arm; calculating the distance between each contact based on the position; and determining the current working state of the target knife switch based on the position, the slope and the distance.

[0005] Optionally, in a first implementation method of the first aspect of the present invention, the use of a pre-trained convolutional neural network to extract the structural edge point cloud data of the target knife gate in the point cloud image and construct a three-dimensional image of the target knife gate includes: using a pre-trained convolutional neural network to obtain the curvature value of each point in the point cloud image; extracting all point clouds whose curvature values ​​are within a pre-defined curvature value threshold of the structural edge point cloud data of the target knife gate, and generating the edge structure of the target knife gate based on the point cloud using the convolutional neural network; and constructing a three-dimensional image of the target knife gate based on the structural edge point cloud data of the target knife gate and the edge structure.

[0006] Optionally, in a second implementation method of the first aspect of the present invention, the three-dimensional image of the target knife switch is constructed based on the structural edge point cloud data of the target knife switch and the edge structure, including: extracting the point cloud feature vectors of each contact and the point cloud feature vectors of each knife arm contained in the preset knife switch; classifying the point cloud vectors of the knife switch based on the point cloud feature vectors of each contact and the point cloud feature vectors of each knife arm, and determining the point cloud set of each contact and the point cloud set of each knife arm; sorting the point cloud set of each contact and the point cloud set of each knife arm based on the sorting specification set in the edge structure; respectively encapsulating the point cloud set of each contact and the point cloud set of each knife arm after the sorting is completed, and texture processing the encapsulated point cloud set to generate a three-dimensional image of the target knife switch.

[0007] Optionally, in a third implementation of the first aspect of the present invention, before using a pre-trained convolutional neural network to extract the structural edge point cloud data of the target knife switch in the point cloud image and construct a three-dimensional image of the target knife switch, it also includes: acquiring the structural edge point cloud data of the target knife switch in different forms multiple times to generate an original data set; using an extraction network to extract from the data set a first sub-data set corresponding to knife switches of different forms and a second sub-data set corresponding to different stages of the opening and closing process of the target knife switch; inputting the first sub-data set and the second sub-data set into convolutional neural networks of each level for training to obtain a convolutional neural network capable of recognizing different forms of the target knife switch and the opening and closing actions of the target knife switch.

[0008] Optionally, in a fourth implementation of the first aspect of the present invention, the position of each contact and each blade arm of the target knife switch in the three-dimensional image, as well as the slope of each blade arm, is determined by using a straight line segment detection method, including: projecting the three-dimensional image onto a two-dimensional plane according to spatial density, and outputting the image to be identified of the target knife switch; identifying each contact and each blade arm in the image to be identified by using image segmentation technology; and detecting the straight line contour of each contact and the straight line contour of each blade arm in the image to be identified by using a straight line segment detection method, generating a contact straight line segment set and A set of knife arm straight segments; respectively obtaining the coordinates of the contact straight segment set and the knife arm straight segment set in the image to be identified; extracting the endpoint coordinates of the longest line segment in the knife arm straight segment set, and calculating the slope of each knife arm based on the endpoint coordinates of the longest line segment; determining the three-dimensional coordinates of the coordinates of the contact straight segment set and the coordinates of the knife arm straight segment set in the three-dimensional image according to the mapping relationship between the image to be identified and the three-dimensional image; and outputting the three-dimensional coordinates as the positions of each contact and each knife arm of the target knife switch in the three-dimensional image.

[0009] Optionally, in the fifth implementation method of the first aspect of the present invention, after using the straight line segment detection method to determine the positions of each contact and each knife arm of the target knife switch in the three-dimensional image, as well as the slope of each knife arm, it also includes: extracting the coordinates of the edge straight line of the knife arm in the three-dimensional image when the knife switch is open or closed; drawing a labeling box for the edge straight line based on the coordinates to mark the position of the edge straight line.

[0010] Optionally, in a sixth implementation method of the first aspect of the present invention, the current working state of the target knife gate is determined based on the position, the slope and the distance, including: comparing the slope and the distance with the slope value threshold and the distance threshold in the standard open or closed state to obtain a first comparison result; comparing the position with the position of the marked edge straight line to obtain a second comparison result; if the first comparison result does not meet the threshold of the relevant information in the standard open or closed state, or the second comparison result does not meet the position of the edge straight line marked in the open or closed state of the knife gate, it is determined that the knife gate is in a state of not being fully opened or not being fully closed; if the first comparison result meets the threshold of the relevant information in the standard open or closed state, and the second comparison result meets the position of the edge straight line marked in the open or closed state of the knife gate, it is determined that the knife gate is in a state of being fully opened or fully closed.

[0011] The second aspect of the present invention provides a device for identifying the open and closed states of a knife switch, comprising: a receiving module for transmitting a detection signal to a target area using a laser radar and receiving a reflected echo signal; a generating module for comparing and analyzing the echo signal and the detection signal, extracting point cloud feature parameters related to the target area, and generating a point cloud image of the target area based on the point cloud feature parameters; an extracting module for extracting structural edge point cloud data of the target knife switch in the point cloud image using a pre-trained convolutional neural network, and constructing a three-dimensional image of the target knife switch; a first determining module for determining the positions of each contact and each knife arm of the target knife switch in the three-dimensional image, as well as the slope of each knife arm, using a straight line segment detection method; a calculating module for calculating the distance between each contact based on the position; and a second determining module for determining the current working state of the target knife switch based on the position, the slope and the distance.

[0012] Optionally, in a first implementation manner of the second aspect of the present invention, the extraction module includes: an acquisition unit, used to obtain the curvature value of each point in the point cloud image using a pre-trained convolutional neural network; a generation unit, used to extract all point clouds whose curvature values ​​are within a pre-defined curvature value threshold of the structural edge point cloud data of the target knife gate, and generate the edge structure of the target knife gate based on the point cloud using the convolutional neural network; a construction unit, used to construct a three-dimensional image of the target knife gate based on the structural edge point cloud data of the target knife gate and the edge structure.

[0013] Optionally, in a second implementation of the second aspect of the present invention, the construction unit includes: an extraction subunit, used to extract the point cloud feature vectors of each contact and the point cloud feature vectors of each knife arm contained in a preset knife switch; a classification subunit, used to classify the point cloud vectors of the knife switch based on the point cloud feature vectors of each contact and the point cloud feature vectors of each knife arm, and determine the point cloud set of each contact and the point cloud set of each knife arm; a sorting unit, used to sort the point cloud set of each contact and the point cloud set of each knife arm based on the sorting specification set in the edge structure; a processing subunit, used to respectively encapsulate the point cloud set of each contact and the point cloud set of each knife arm after the sorting is completed, and perform texture processing on each point cloud set after the encapsulation to generate a three-dimensional image of the target knife switch.

[0014] Optionally, in a third implementation of the second aspect of the present invention, the extraction module is further used to: obtain structural edge point cloud data of the target knife switch in different forms multiple times to generate an original data set; use an extraction network to extract from the data set a first sub-data set corresponding to knife switches of different forms and a second sub-data set corresponding to different stages of the target knife switch during the opening and closing process; input the first sub-data set and the second sub-data set into convolutional neural networks of each level for training to obtain a convolutional neural network capable of recognizing different forms of the target knife switch and the opening and closing actions of the target knife switch.

[0015] Optionally, in a fourth implementation of the second aspect of the present invention, the first determination module includes: a first output unit, for projecting the three-dimensional image onto a two-dimensional plane according to spatial density, and outputting the image to be identified of the target knife switch; a first identification unit, for identifying each of the contacts and each of the knife arms in the image to be identified by using image segmentation technology; a second identification unit, for detecting the straight line contours of each of the contacts and the straight line contours of each of the knife arms in the image to be identified by using a straight line segment detection method, and generating a contact straight line segment set and a knife arm straight line segment set; a collection unit, for obtaining the contact straight line segments respectively. The coordinates of the head straight segment set and the knife arm straight segment set in the image to be identified; a calculation unit, used to extract the endpoint coordinates of the longest line segment in the knife arm straight segment set, and calculate the slope of each knife arm based on the endpoint coordinates of the longest line segment; a determination unit, used to determine the three-dimensional coordinates of the coordinates of the contact straight segment set and the coordinates of the knife arm straight segment set in the three-dimensional image according to the mapping relationship between the image to be identified and the three-dimensional image; a second output unit, used to output the three-dimensional coordinates as the positions of each contact and each knife arm of the target knife switch in the three-dimensional image.

[0016] Optionally, in a fifth implementation of the second aspect of the present invention, the first determination module is further used to: extract the coordinates of the edge straight line of the knife arm in the three-dimensional image when the knife gate is opened or closed, respectively; and draw a labeling box for the edge straight line based on the coordinates to mark the position of the edge straight line.

[0017] Optionally, in a sixth implementation manner of the second aspect of the present invention, the second determination module includes: a first comparison unit, for comparing the slope and the distance with a slope value threshold and a distance threshold in a standard open or closed state to obtain a first comparison result; a second comparison unit, for comparing the position with the marked position of the edge straight line to obtain a second comparison result; a first determination unit, for determining that the knife gate is in a state of not being fully opened or not being fully closed if the first comparison result does not meet the threshold of the relevant information in the standard open or closed state, or the second comparison result does not meet the position of the edge straight line marked in the open or closed state of the knife gate; a second determination unit, for determining that the knife gate is in a state of being fully opened or fully closed if the first comparison result meets the threshold of the relevant information in the standard open or closed state, and the second comparison result meets the position of the edge straight line marked in the open or closed state of the knife gate.

[0018] The third aspect of the present invention provides an electronic device, comprising: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected via lines; the at least one processor calls the instructions in the memory so that the electronic device executes the above-mentioned method for identifying the opening and closing states of knife switches.

[0019] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned method for identifying the opening and closing states of knife switches.

[0020] In the technical solution of the present invention, a pre-trained convolutional neural network is used to extract the structural edge point cloud data of the target knife switch from the collected point cloud image of the target area, and a three-dimensional image of the target knife switch is constructed. The straight line segment detection method is used to determine the positions of each contact and each knife arm of the target knife switch in the three-dimensional image, and the slope of each knife arm. The distance between each contact is calculated based on the position, thereby determining the current working state of the target knife switch, and realizing accurate identification of the open and closed state of the knife switch in various environments and pixels. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Schematic diagram of a first embodiment of a method for identifying the opening and closing states of a knife switch according to an embodiment of the present invention;

[0022] Figure 2 Schematic diagram of a second embodiment of a method for identifying the opening and closing states of a knife switch according to an embodiment of the present invention;

[0023] Figure 3 Schematic diagram of a third embodiment of a method for identifying the opening and closing states of a knife switch according to an embodiment of the present invention;

[0024] Figure 4 A schematic diagram of a module framework of a knife switch device according to an embodiment of the present invention;

[0025] Figure 5 A schematic diagram of an embodiment of a device for identifying the opening and closing states of a knife switch according to an embodiment of the present invention;

[0026] Figure 6 Schematic diagram of another embodiment of the device for identifying the opening and closing states of a knife switch according to an embodiment of the present invention;

[0027] Figure 7 FIG. 1 is a schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In response to the existing methods for identifying the opening and closing states of knife switches, the present invention uses a pre-trained convolutional neural network to extract the structural edge point cloud data of the target knife switch from the collected point cloud image of the target area, constructs a three-dimensional image of the target knife switch, and uses a straight line segment detection method to determine the positions of each contact and each knife arm of the target knife switch in the three-dimensional image, as well as the slope of each knife arm. Based on the position, the distance between each contact is calculated, thereby determining the current working state of the target knife switch, and realizing accurate identification of the opening and closing state of the knife switch in various environments and pixels.

[0029] The terms "first," "second," "third," "fourth," and so on (if any) in the description and claims of the present invention and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that shown or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus that includes a series of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, product, or apparatus.

[0030] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 The first embodiment of the method for identifying the switch opening and closing states in the embodiment of the present invention includes:

[0031] 101. Use laser radar to transmit detection signals to the target area and receive reflected echo signals;

[0032] The laser scanner is fixed to the radar swing bracket. The radar swing bracket and the swing control device housing are connected by a rotating bearing. The rotating bearing is fixed to the side wall of the swing control device housing. The laser scanner is controlled to swing to achieve three-dimensional scanning of the knife switch. The laser scanner transmits a detection signal to the target area, and the echo signal receiving end receives the echo signal reflected from the target area.

[0033] 102. Compare and analyze the echo signal and the detection signal, extract point cloud feature parameters related to the target area, and generate a point cloud image of the target area based on the point cloud feature parameters;

[0034] The detection signal and the echo signal are compared and analyzed, and all signals in the echo signal that are different from the detection signal are extracted and output as signals related to the target area. The point cloud parameter information carried in the signal related to the target area is then extracted, and the outliers in the point cloud parameter information are directly eliminated or processed in a smooth manner. The remaining point cloud parameter information is used as a data set of spatial points in the target area. Each point contains three-dimensional coordinate information, color information, reflection intensity information and echo number information, and a point cloud image of the target area is generated based on the remaining point cloud parameter information.

[0035] 103. Use the pre-trained convolutional neural network to extract the structural edge point cloud data of the target knife switch in the point cloud image and construct a three-dimensional image of the target knife switch;

[0036] A pre-trained convolutional neural network is used to segment the point cloud image of the target knife switch from the point cloud image of the target area, and the point cloud data of the target knife switch is extracted based on the point cloud image of the target knife switch, and the acquired point cloud data of the target knife switch is processed to obtain point cloud distribution data; the obtained point cloud distribution data is processed to obtain associated feature distance data and point cloud vector data of the corresponding point cloud in a preset neighborhood; point cloud similarity data is obtained based on the obtained associated feature distance data; the point cloud vector data is classified based on the obtained point cloud similarity data to obtain edge features of the point cloud corresponding to the point cloud data in the preset neighborhood, and the edge features are output as structural edge point cloud data of the target knife switch.

[0037] The structural edge point cloud data of the target knife switch in different forms are obtained multiple times to generate an original data set. The extraction network is used to extract the first sub-dataset corresponding to the knife switches in different forms, the second sub-dataset corresponding to the different stages of the target knife switch in the opening and closing process, and the third sub-dataset corresponding to the moving contact or the static contact from the data set. The first sub-dataset, the second sub-dataset, and the third sub-dataset are input into the convolutional neural network of each level for training, and a convolutional neural network with the ability to recognize the different forms of the target knife switch, the opening and closing actions of the target knife switch, and the moving contact and the static contact is obtained.

[0038] Furthermore, based on the 3D spatial map constructed from the 3D point cloud data acquired by the LiDAR, convolutional neural network technology was used to annotate and train the knife switches in the 3D spatial map, enabling the algorithm to recognize various shapes in the 3D spatial map. Similarly, convolutional neural networks were used to train and learn the movement of knife switches of various shapes, as well as their static and moving contacts, enabling the algorithm to recognize the opening and closing movements of knife switches of various shapes, as well as the static and moving contacts.

[0039] A Laplacian pyramid is constructed based on the structural edge point cloud data of the target knife switch, and the Laplacian pyramid is fused and calculated to obtain a three-dimensional image of the target knife switch.

[0040] A three-dimensional image is mapped into a two-dimensional image from different angles to obtain multiple images. The fusion coefficient of each pixel in the multiple images is calculated respectively, and the fusion coefficient Gaussian pyramid of the two images is calculated respectively based on the fusion coefficient of each pixel in the two images; the Laplacian pyramid of the two images is calculated respectively; the two images are fused according to the fusion coefficient Gaussian pyramid and the Laplacian pyramid of the two images to obtain a fused Laplacian pyramid; the fused image is reconstructed based on the fused Laplacian pyramid and output as a three-dimensional image of the target knife switch.

[0041] After constructing the three-dimensional image, the two-dimensional image of the knife switch is acquired and saved in real time from a monitoring device located on the knife switch, and the front view, side view and top view of the three-dimensional image of the target knife switch are extracted.

[0042] 104. Determine the position of each contact and each blade arm of the target switch in the three-dimensional image, as well as the slope of each blade arm, using a straight line segment detection method;

[0043] The three-dimensional image is projected onto a two-dimensional plane according to spatial density, and the output is an image to be identified of the target knife switch; the image segmentation technology is used to identify each contact and each knife arm in the image to be identified; the straight line profile of each contact and each knife arm in the image to be identified is detected respectively by the straight line segment detection method, and a contact straight line segment set and a knife arm straight line segment set are generated; the coordinates of the contact straight line segment set and the knife arm straight line segment set in the image to be identified are obtained respectively; the endpoint coordinates of the longest line segment in the knife arm straight line segment set are extracted, and the slope of each knife arm is calculated based on the endpoint coordinates of the longest line segment; according to the mapping relationship between the image to be identified and the three-dimensional image, the three-dimensional coordinates of the coordinates of the contact straight line segment set and the coordinates of the knife arm straight line segment set in the three-dimensional image are determined; the three-dimensional coordinates are output as the positions of each contact and each knife arm of the target knife switch in the three-dimensional image.

[0044] 105. Calculate the distance between each contact based on the position;

[0045] Identify each contact in the 3D image and distinguish them based on their shape differences. Calculate the distances between the moving contacts and between the moving contacts and the stationary contacts using a three-dimensional distance calculation formula.

[0046] 106. Determine the current working status of the target switch based on the position, slope and distance.

[0047] The backend of the switch opening and closing identification system supports northbound connection, sending the corresponding switch opening and closing signals to the dispatching system and other systems. After determining the current operating status of the target switch based on location, slope, and distance, if the operating status is open or closed, a closed or open signal is generated and output to the measurement and control device.

[0048] In an embodiment of the present invention, a pre-trained convolutional neural network is used to extract structural edge point cloud data of a target knife switch from a collected point cloud image of a target area, and a three-dimensional image of the target knife switch is constructed. The position of each contact and each knife arm of the target knife switch in the three-dimensional image, as well as the slope of each knife arm, is determined by a straight line segment detection method. The distance between each contact is calculated based on the position, thereby determining the current working state of the target knife switch, and realizing accurate recognition of the open and closed state of the knife switch in various environments and pixels.

[0049] See also Figure 2 The second embodiment of the method for identifying the switch opening and closing state in the embodiment of the present invention includes the following steps:

[0050] 201. Use laser radar to transmit detection signals to the target area and receive reflected echo signals;

[0051] The corner points and surface points of the knife switch are extracted, and the criticality of each sub-area on the knife switch is evaluated using a custom factor. The particle filter method is used to fuse the information of the current frame and the historical frame to ensure that a stable and unchanging target area is selected.

[0052] 202. Compare and analyze the echo signal and the detection signal, extract point cloud feature parameters related to the target area, and generate a point cloud image of the target area based on the point cloud feature parameters;

[0053] 203. Using a pre-trained convolutional neural network to obtain the curvature value of each point in the point cloud image;

[0054] A convolutional neural network is used to obtain the feature points of multiple structures in the point cloud image, and curve fitting is performed on each feature point to obtain the trajectory reference line of each structure in the point cloud image. The curvature of the trajectory reference line is identified, and the degree of bending of each structure is determined based on the curvature, thereby obtaining the trajectory of the structure contained in the point cloud image.

[0055] 204. Extract all point clouds whose curvature values ​​are within a predefined curvature value threshold of the structure edge point cloud data of the target knife switch, and generate the edge structure of the target knife switch using a convolutional neural network based on the point clouds;

[0056] A curvature threshold for the target switch's structural edge point cloud data is predefined and extracted from the point cloud image based on the curvature threshold. A convolutional neural network is then used to generate the target switch's edge structure based on the target switch's point cloud, following the method used to obtain the structural trajectory.

[0057] 205. Construct a three-dimensional image of the target knife switch based on the structural edge point cloud data and edge structure of the target knife switch;

[0058] Extract the point cloud feature vectors of each contact and each knife arm contained in the preset knife switch; classify the point cloud vectors of the knife switch based on the point cloud feature vectors of each contact and the point cloud feature vectors of each knife arm, and determine the point cloud set of each contact and the point cloud set of each knife arm; sort the point cloud set of each contact and the point cloud set of each knife arm based on the sorting specification set in the edge structure; respectively encapsulate the point cloud set of each contact and the point cloud set of each knife arm after sorting, and perform texture processing on the encapsulated point cloud sets to generate a three-dimensional image of the target knife switch.

[0059] 206. Determine the position of each contact and each blade arm of the target switch in the three-dimensional image, as well as the slope of each blade arm, using a straight line segment detection method;

[0060] After using the straight line detection method to determine the positions of each contact and each blade arm of the target knife switch in the three-dimensional image, as well as the slope of each blade arm, the coordinates of the edge lines of the blade arms in the three-dimensional image are extracted when the knife switch is open or closed. Based on the coordinates, a labeling box for the edge line is drawn to mark the position of the edge line.

[0061] 207. Calculate the distance between each contact based on the position;

[0062] 208. Determine the current working status of the target knife switch based on the position, slope and distance.

[0063] If the working status is that the switch is not in place or the switch is not in place, the switch is identified for two consecutive times. If the results of the three times are all that the switch is not in place or the switch is not in place, an alarm signal is sent to the measurement and control device.

[0064] Obtain the content of location information, slope information and distance information; based on the content, generate alarm information according to the preset complaint rules; based on the alarm information, call the corresponding alarm process to make a complaint.

[0065] In an embodiment of the present invention, the structural edge point cloud data of the target knife switch is extracted from the point cloud image based on the curvature threshold of the target knife switch, and a three-dimensional image of the target knife switch is established. The position of each contact and each knife arm in the three-dimensional image, the slope of each knife arm and the distance between each contact are calculated, and the position, slope and distance of the target knife switch are detected, so as to achieve the purpose of judging whether the knife switch is opened and closed in place.

[0066] See also Figure 3 The third embodiment of the method for identifying the switch opening and closing states in the embodiment of the present invention is applied to Figure 4 In the provided knife switch device, the knife switch device includes contacts 401 and 404, and knife arms 402 and 403. The knife switch open / close state identification includes the following steps:

[0067] 301. Use laser radar to transmit detection signals to the target area and receive reflected echo signals;

[0068] 302. Compare and analyze the echo signal and the detection signal, extract point cloud feature parameters related to the target area, and generate a point cloud image of the target area based on the point cloud feature parameters;

[0069] 303. Using a pre-trained convolutional neural network to extract the structural edge point cloud data of the target knife switch in the point cloud image, and construct a three-dimensional image of the target knife switch;

[0070] An untrained convolutional neural network module is constructed and initialized. Several point clouds of knife switches in different forms are input into the initialized untrained convolutional neural network module to extract feature maps. Based on the known feature maps, the commonly activated neuron clusters of each category of images are extracted and saved. The point cloud image of the target area is input into the initialized untrained convolutional neural network module to extract the feature maps of the image to be identified. Based on the feature maps, the activated neuron clusters of the image to be identified are extracted. The activated neuron clusters of the image to be identified are compared one by one with the commonly activated neuron clusters of each category of images that have been saved, and the structural edge point cloud data of the target knife switch is determined from the point cloud image of the target area.

[0071] 304. Project the three-dimensional image onto a two-dimensional plane according to spatial density, and output it as an image to be identified of the target knife switch;

[0072] Establishing a transformation relationship between a point cloud coordinate system and an image coordinate system, the transformation relationship including transformation parameters; back-projecting the three-dimensional point cloud onto the plane where the two-dimensional image is located using the transformation relationship to obtain a projection point set of the three-dimensional point cloud; clustering the projection point set to obtain a cluster center position value of the projection point set; and performing edge detection on the two-dimensional image to obtain an image to be identified;

[0073] 305. Identify each contact and each blade arm in the image to be identified using image segmentation technology;

[0074] Before recognition, the two-dimensional projection image is morphologically processed to obtain a denoised image to be recognized. An image segmentation model is constructed, based on which each contact and blade arm in the image to be recognized is segmented according to the standard contact and blade arm. The segmentation lines corresponding to the characteristics or information of each contact and blade arm are determined to obtain the image of each contact and blade arm.

[0075] 306. Detect the straight line profile of each contact and the straight line profile of each blade arm in the image to be identified using a straight line segment detection method, and generate a contact straight line segment set and a blade arm straight line segment set;

[0076] Based on the straight line segment detection method, detection is performed in the contact image and the knife arm image, edge detection is performed on the contact image, and all straight line segments in the contact image are extracted to form a contact straight line segment set; edge detection is performed on the knife arm image, and all straight line segments in the knife arm image are extracted to form a knife arm straight line segment set.

[0077] 307. Obtain the coordinates of the contact straight line segment set and the knife arm straight line segment set in the image to be identified respectively;

[0078] The relative positions of each line segment in the contact straight line segment set within the image to be identified are obtained, and the coordinates of each line segment in the straight line segment set within the image to be identified are determined based on the relative positions. Similarly, the coordinates of each line segment in the tool arm straight line segment set within the image to be identified are determined based on the relative positions of each line segment in the tool arm straight line segment set within the image to be identified.

[0079] 308. Extract the endpoint coordinates of the longest line segment in the set of straight line segments of the blade arm, and calculate the slope of each blade arm based on the endpoint coordinates of the longest line segment;

[0080] The endpoint coordinates of the longest line segment in the set of straight line segments of the knife arm are extracted, the slope of the longest line segment is calculated based on the coordinates, the slope of the line is calculated based on the slope, and the slope is output as the slope of the knife arm.

[0081] After calculating the slope of each blade arm, mark the edge straight line of the blade arm in the open or closed state.

[0082] 309. Determine the three-dimensional coordinates of the contact straight line segment set and the blade arm straight line segment set in the three-dimensional image based on the mapping relationship between the image to be recognized and the three-dimensional image;

[0083] Extract feature points of the image to be identified, obtain the mapping relationship between the image to be identified and the three-dimensional image by learning the mapping relationship between the original two-dimensional image feature points and the original three-dimensional image feature points in the training sample, use this mapping relationship to estimate the feature points of the image to be identified to obtain the three-dimensional image feature points corresponding to the two-dimensional image feature points, and verify the mapping relationship between the image to be identified and the three-dimensional image based on the feature points of the three-dimensional image. If the verification is correct, input the coordinates of the contact straight line segment set and the coordinates of the knife arm straight line segment set into the mapping relationship respectively to obtain the coordinates of the contact straight line segments and the knife arm straight line segments in the three-dimensional image.

[0084] 310. Output the three-dimensional coordinates as the positions of each contact and each blade arm of the target switch in the three-dimensional image, and calculate the distance between each contact based on the positions;

[0085] A key point is selected in each contact, and the coordinates of each key point in the three-dimensional image are extracted respectively. The distance between each contact is calculated based on the coordinates of each key point.

[0086] The distance between contacts 401 and 404 is measured when the switch is in the standard open or standard closed state, and the output is the distance threshold for the standard open or closed state. The longest line segments on arms 402 and 403 are annotated in the 3D graph. The points on the longest line segments of the two switch arms, away from the contacts, are also marked. This connecting line serves as the annotated edge line. The slope of the arm is calculated based on the slope of the longest line segments on arms 402 and 403.

[0087] 311. Determine the current working status of the target knife switch based on position, slope and distance.

[0088] After determining the current working state of the target knife switch based on the position, slope and distance, obtain the longest line segment on the knife arms 402 and 403 away from the points of the contacts 401 and 404 and connect them, and use this line segment as the marked edge straight line.

[0089] Compare the slope and distance with the slope value threshold and distance threshold in the standard open or closed state to obtain a first comparison result; compare the position with the position of the marked edge straight line to obtain a second comparison result; if the first comparison result does not meet the threshold of the relevant information in the standard open or closed state, or the second comparison result does not meet the position of the edge straight line marked when the knife gate is open or closed, it is determined that the knife gate is in a state of not being fully opened or not being fully closed; if the first comparison result meets the threshold of the relevant information in the standard open or closed state, and the second comparison result meets the position of the edge straight line marked when the knife gate is open or closed, it is determined that the knife gate is in a state of being fully opened or fully closed.

[0090] In an embodiment of the present invention, a three-dimensional image of a target knife switch is established and projected onto a two-dimensional plane, and the contacts and blade arms of the target knife switch are identified based on the obtained plane image. A straight line segment detection method is used to generate a contact straight line segment set and a blade arm straight line segment set, and the coordinates of the contact straight line segment set and the blade arm straight line segment set in the image to be identified are obtained. Based on the coordinates, the slope of each blade arm and the distance between each contact are calculated to detect the position, slope and distance of the target knife switch, thereby achieving the purpose of judging whether the knife switch is opened and closed in place.

[0091] The above describes the method for identifying the switch opening and closing states in the embodiment of the present invention. The following describes the device for identifying the switch opening and closing states in the embodiment of the present invention. Figure 4 In one embodiment of the present invention, a device for identifying the opening and closing states of a knife switch includes:

[0092] The receiving module 501 is used to transmit a detection signal to the target area using a laser radar and receive a reflected echo signal;

[0093] a generating module 502 for comparing and analyzing the echo signal and the detection signal, extracting point cloud feature parameters related to the target area, and generating a point cloud image of the target area based on the point cloud feature parameters;

[0094] An extraction module 503 is configured to extract structural edge point cloud data of the target knife switch from the point cloud image using a pre-trained convolutional neural network, and construct a three-dimensional image of the target knife switch;

[0095] A first determining module 504 is configured to determine the position of each contact and each blade arm of the target knife switch in the three-dimensional image, as well as the slope of each blade arm, using a straight line segment detection method;

[0096] A calculation module 505, configured to calculate the distance between the contacts based on the positions;

[0097] The second determining module 506 is configured to determine the current working state of the target knife switch based on the position, the slope and the distance.

[0098] In an embodiment of the present invention, a pre-trained convolutional neural network is used to extract structural edge point cloud data of a target knife switch from a collected point cloud image of a target area, and a three-dimensional image of the target knife switch is constructed. The position of each contact and each knife arm of the target knife switch in the three-dimensional image, as well as the slope of each knife arm, is determined by a straight line segment detection method. The distance between each contact is calculated based on the position, thereby determining the current working state of the target knife switch, and realizing accurate recognition of the open and closed state of the knife switch in various environments and pixels.

[0099] See also Figure 5 Another embodiment of the device for identifying the opening and closing states of a knife switch according to the present invention includes:

[0100] The receiving module 501 is used to transmit a detection signal to the target area using a laser radar and receive a reflected echo signal;

[0101] a generating module 502 for comparing and analyzing the echo signal and the detection signal, extracting point cloud feature parameters related to the target area, and generating a point cloud image of the target area based on the point cloud feature parameters;

[0102] An extraction module 503 is configured to extract structural edge point cloud data of the target knife switch from the point cloud image using a pre-trained convolutional neural network, and construct a three-dimensional image of the target knife switch;

[0103] A first determining module 504 is configured to determine the position of each contact and each blade arm of the target knife switch in the three-dimensional image, as well as the slope of each blade arm, using a straight line segment detection method;

[0104] A calculation module 505, configured to calculate the distance between the contacts based on the positions;

[0105] The second determining module 506 is configured to determine the current working state of the target knife switch based on the position, the slope and the distance.

[0106] In this embodiment, the extraction module 503 includes:

[0107] an acquisition unit 5031, configured to acquire a curvature value of each point in the point cloud image using a pre-trained convolutional neural network;

[0108] A generating unit 5032 is configured to extract all point clouds whose curvature values ​​are within a predefined curvature value threshold of the structure edge point cloud data of the target knife gate, and generate the edge structure of the target knife gate based on the point clouds using the convolutional neural network;

[0109] The construction unit 5033 is configured to construct a three-dimensional image of the target knife gate based on the structural edge point cloud data of the target knife gate and the edge structure.

[0110] In this embodiment, the construction unit 5033 includes:

[0111] The extraction subunit 50331 is used to extract the point cloud feature vectors of each contact and each knife arm contained in the preset knife switch;

[0112] The classification subunit 50332 is configured to classify the point cloud vectors of the knife switch based on the point cloud feature vectors of the contacts and the point cloud feature vectors of the knife arms, and determine the point cloud set of the contacts and the point cloud set of the knife arms;

[0113] a sorting subunit 50333, configured to sort the point cloud sets of the contacts and the point cloud sets of the blade arms based on a sorting specification set in the edge structure;

[0114] The processing sub-unit 50334 is used to respectively encapsulate the point cloud sets of each contact and each knife arm after sorting, and perform texture processing on each encapsulated point cloud set to generate a three-dimensional image of the target knife switch.

[0115] In this embodiment, the extraction module 503 is further configured to:

[0116] The structural edge point cloud data of the target knife switch in different forms are obtained multiple times to generate an original data set; the first sub-data set corresponding to the knife switches in different forms and the second sub-data set corresponding to the target knife switch at different stages in the opening and closing process are extracted from the data set using an extraction network; the first sub-data set and the second sub-data set are input into convolutional neural networks at each level for training, thereby obtaining a convolutional neural network capable of recognizing the different forms of the target knife switch and the opening and closing actions of the target knife switch.

[0117] In this embodiment, the first determining module 504 includes:

[0118] The first output unit 5041 is configured to project the three-dimensional image onto a two-dimensional plane according to spatial density, and output the projected image as the target knife switch image to be identified.

[0119] A first recognition unit 5042 is configured to recognize each contact and each blade arm in the image to be recognized by using image segmentation technology;

[0120] The second recognition unit 5043 is configured to detect the straight line profiles of the contacts and the straight line profiles of the blade arms in the image to be recognized using a straight line segment detection method, and generate a contact straight line segment set and a blade arm straight line segment set;

[0121] A collection unit 5044 is used to respectively obtain the coordinates of the contact straight line segment set and the knife arm straight line segment set in the image to be identified;

[0122] a calculation unit 5045, configured to extract the endpoint coordinates of the longest line segment in the set of straight line segments of the knife arm, and calculate the slope of each knife arm based on the endpoint coordinates of the longest line segment;

[0123] a determining unit 5046 configured to determine the three-dimensional coordinates of the coordinates of the set of contact straight segments and the set of blade arm straight segments in the three-dimensional image based on a mapping relationship between the image to be identified and the three-dimensional image;

[0124] The second output unit 5047 is used to output the three-dimensional coordinates as the positions of each contact and each knife arm of the target knife switch in the three-dimensional image.

[0125] In this embodiment, the first determining module 504 is further configured to:

[0126] The coordinates of the edge line of the knife arm in the three-dimensional image are extracted respectively when the knife gate is opened or closed; and a marking frame of the edge line is drawn based on the coordinates to mark the position of the edge line.

[0127] In this embodiment, the second determining module 506 includes:

[0128] A first comparison unit 5061 is configured to compare the slope and the distance with a slope value threshold and a distance threshold in a standard open or closed state to obtain a first comparison result;

[0129] A second comparison unit 5062 is configured to compare the position with the marked position of the edge line to obtain a second comparison result;

[0130] A first determining unit 5063 is configured to determine that the knife gate is in a partially opened or partially closed state if the first comparison result does not meet a threshold of relevant information in a standard open or closed state, or if the second comparison result does not match the position of the edge straight line marked in the open or closed state of the knife gate;

[0131] The second determination unit 5064 is used to determine whether the knife gate is in the open or closed state if the first comparison result meets the threshold of the relevant information in the standard open or closed state, and the second comparison result is consistent with the position of the edge straight line marked in the open or closed state of the knife gate.

[0132] In an embodiment of the present invention, a pre-trained convolutional neural network is used to extract structural edge point cloud data of a target knife switch from a collected point cloud image of a target area, and a three-dimensional image of the target knife switch is constructed. The position of each contact and each knife arm of the target knife switch in the three-dimensional image, as well as the slope of each knife arm, is determined by a straight line segment detection method. The distance between each contact is calculated based on the position, thereby determining the current working state of the target knife switch, and realizing accurate recognition of the open and closed state of the knife switch in various environments and pixels.

[0133] above Figure 5 and Figure 6 The knife switch opening and closing state identification device in the embodiment of the present invention is described in detail from the perspective of modular functional entities, and the electronic device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0134] Figure 7is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. The electronic device 700 may vary significantly due to different configurations or performance, and may include one or more processors (central processing units, CPUs) 710 (for example, one or more processors), a memory 720, and one or more storage media 730 (for example, one or more mass storage devices) storing application programs 733 or data 732. The memory 720 and storage medium 730 may be either transient or persistent storage. The program stored in the storage medium 730 may include one or more modules (not shown), each of which may include a series of instruction operations on the electronic device 700. Furthermore, the processor 710 may be configured to communicate with the storage medium 730 to execute the series of instruction operations in the storage medium 730 on the electronic device 700.

[0135] The electronic device 700 may further include one or more power supplies 740, one or more wired or wireless network interfaces 750, one or more input and output interfaces 760, and / or one or more operating systems 731, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 7 The illustrated electronic device structure does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0136] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, cause the computer to execute the steps of the knife switch opening and closing state identification method.

[0137] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0138] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0139] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying the opening and closing states of a knife switch, characterized in that: The method for identifying the switch opening and closing states includes: Use laser radar to transmit detection signals to the target area and receive reflected echo signals; Comparing and analyzing the echo signal and the detection signal, extracting point cloud feature parameters related to the target area, and generating a point cloud image of the target area based on the point cloud feature parameters; Extracting structural edge point cloud data of the target knife switch from the point cloud image using a pre-trained convolutional neural network to construct a three-dimensional image of the target knife switch; Determine the position of each contact and each blade arm of the target knife switch in the three-dimensional image and the slope of each blade arm by using a straight line segment detection method; calculating the distance between each of the contacts based on the positions; Determine the current working state of the target knife switch based on the position, the slope and the distance; The method of extracting structural edge point cloud data of the target knife switch in the point cloud image using a pre-trained convolutional neural network to construct a three-dimensional image of the target knife switch includes: Using a pre-trained convolutional neural network to obtain the curvature value of each point in the point cloud image; Extracting all point clouds whose curvature values ​​are within a predefined curvature value threshold of the structure edge point cloud data of the target knife gate, and generating the edge structure of the target knife gate based on the point clouds using the convolutional neural network; Constructing a three-dimensional image of the target knife gate based on the structural edge point cloud data of the target knife gate and the edge structure; The step of constructing a three-dimensional image of the target knife gate based on the structural edge point cloud data of the target knife gate and the edge structure includes: Extracting the point cloud feature vectors of each contact and each knife arm contained in the preset knife switch; Classifying the point cloud vectors of the knife switch based on the point cloud feature vectors of each contact and the point cloud feature vectors of each knife arm, and determining the point cloud set of each contact and the point cloud set of each knife arm; sorting the point cloud sets of each of the contacts and the point cloud sets of each of the blade arms based on a sorting specification set in the edge structure; The point cloud sets of the contacts and the point cloud sets of the knife arms after sorting are respectively packaged, and texture processing is performed on the packaged point cloud sets to generate a three-dimensional image of the target knife switch.

2. The method for identifying the switch opening and closing state according to claim 1, characterized in that: Before extracting the structural edge point cloud data of the target knife switch in the point cloud image using the pre-trained convolutional neural network to construct the three-dimensional image of the target knife switch, the method further includes: Acquire structural edge point cloud data of the target knife switch in different forms multiple times to generate an original data set; Extracting first sub-datasets corresponding to knife switches of different shapes and second sub-datasets corresponding to different stages of the target knife switch during the opening and closing process from the data set using an extraction network; The first sub-dataset and the second sub-dataset are input into convolutional neural networks of various levels for training, so as to obtain a convolutional neural network capable of recognizing different forms of the target knife switch and the opening and closing actions of the target knife switch.

3. The method for identifying the switch opening and closing state according to claim 1, characterized in that: The method of using a straight line segment detection method to determine the position of each contact and each blade arm of the target knife switch in the three-dimensional image, as well as the slope of each blade arm, includes: Projecting the three-dimensional image onto a two-dimensional plane according to spatial density, and outputting the image to be identified of the target knife switch; Identifying each of the contacts and each of the blade arms in the image to be identified using image segmentation technology; Using a straight line segment detection method, the straight line profile of each contact and the straight line profile of each knife arm in the image to be identified are respectively detected to generate a contact straight line segment set and a knife arm straight line segment set; Respectively obtaining the coordinates of the contact straight line segment set and the knife arm straight line segment set in the image to be identified; Extracting the endpoint coordinates of the longest line segment in the set of straight line segments of the knife arm, and calculating the slope of each knife arm based on the endpoint coordinates of the longest line segment; determining the three-dimensional coordinates of the coordinates of the set of contact straight segments and the set of blade arm straight segments in the three-dimensional image according to a mapping relationship between the image to be recognized and the three-dimensional image; The three-dimensional coordinates are output as positions of each contact and each knife arm of the target knife switch in the three-dimensional image.

4. The method for identifying the switch opening and closing state according to claim 3 is characterized in that: After determining the positions of the contacts and the blade arms of the target knife switch in the three-dimensional image and the slopes of the blade arms by using the straight line segment detection method, the method further includes: extracting the coordinates of the edge straight line of the knife arm in the three-dimensional image when the knife gate is open or closed; A marking frame of the edge line is drawn based on the coordinates to mark the position of the edge line.

5. The method for identifying the switch opening and closing state according to claim 4, characterized in that: The determining the current working state of the target knife switch based on the position, the slope and the distance includes: Comparing the slope and the distance with a slope value threshold and a distance threshold in a standard open or closed state to obtain a first comparison result; Comparing the position with the marked position of the edge straight line to obtain a second comparison result; If the first comparison result does not meet the threshold of the relevant information in the standard open or closed state, or the second comparison result does not conform to the position of the edge straight line marked in the open or closed state of the knife gate, it is determined that the knife gate is in a state of partially open or partially closed; If the first comparison result meets the threshold of relevant information in the standard open or closed state, and the second comparison result matches the position of the edge straight line marked in the open or closed state of the knife gate, it is determined that the knife gate is in the open or closed state.

6. A device for identifying the opening and closing state of a knife switch, comprising: The receiving module is used to transmit detection signals to the target area using the laser radar and receive the reflected echo signals; a generating module, configured to compare and analyze the echo signal and the detection signal, extract point cloud feature parameters related to the target area, and generate a point cloud image of the target area based on the point cloud feature parameters; An extraction module is used to extract structural edge point cloud data of the target knife switch in the point cloud image using a pre-trained convolutional neural network to construct a three-dimensional image of the target knife switch; a first determining module, configured to determine the position of each contact and each blade arm of the target knife switch in the three-dimensional image, and the slope of each blade arm, by using a straight line segment detection method; a calculation module, configured to calculate the distance between the contacts based on the positions; A second determining module is used to determine the current working state of the target knife switch based on the position, the slope and the distance; The extraction module includes: an acquisition unit for acquiring a curvature value of each point in the point cloud image using a pre-trained convolutional neural network; a generation unit for extracting all point clouds whose curvature values ​​are within a pre-defined curvature value threshold of the structural edge point cloud data of the target knife gate, and generating the edge structure of the target knife gate based on the point clouds using the convolutional neural network; and a construction unit for constructing a three-dimensional image of the target knife gate based on the structural edge point cloud data of the target knife gate and the edge structure. The construction unit includes: an extraction subunit, which is used to extract the point cloud feature vectors of each contact and the point cloud feature vectors of each knife arm contained in a preset knife switch; a classification subunit, which is used to classify the point cloud vectors of the knife switch based on the point cloud feature vectors of each contact and the point cloud feature vectors of each knife arm, and determine the point cloud set of each contact and the point cloud set of each knife arm; a sorting unit, which is used to sort the point cloud set of each contact and the point cloud set of each knife arm based on the sorting specification set in the edge structure; a processing subunit, which is used to respectively encapsulate the point cloud set of each contact and the point cloud set of each knife arm after the sorting is completed, and perform texture processing on each point cloud set after the encapsulation processing to generate a three-dimensional image of the target knife switch.

7. The device for identifying the switch opening and closing state according to claim 6, characterized in that: The extraction module is also used to: obtain the structural edge point cloud data of the target knife switch in different forms multiple times to generate an original data set; use the extraction network to extract the first sub-data set corresponding to the knife switches in different forms and the second sub-data set corresponding to the target knife switch at different stages in the opening and closing process from the data set; input the first sub-data set and the second sub-data set into the convolutional neural network of each level for training, and obtain a convolutional neural network capable of recognizing the different forms of the target knife switch and the opening and closing actions of the target knife switch.

8. The device for identifying the switch opening and closing state according to claim 6, characterized in that: The first determination module includes: a first output unit, used to project the three-dimensional image onto a two-dimensional plane according to spatial density, and output it as the image to be identified of the target knife switch; a first recognition unit, used to use image segmentation technology to identify each of the contacts and each of the knife arms in the image to be identified; a second recognition unit, used to use a straight line segment detection method to respectively detect the straight line contours of each of the contacts and the straight line contours of each of the knife arms in the image to be identified, and generate a contact straight line segment set and a knife arm straight line segment set; a collection unit, used to respectively obtain the coordinates of the contact straight line segment set and the knife arm straight line segment set in the image to be identified; a calculation unit, used to extract the endpoint coordinates of the longest line segment in the knife arm straight line segment set, and calculate the slope of each knife arm based on the endpoint coordinates of the longest line segment; a determination unit, used to determine the three-dimensional coordinates of the coordinates of the contact straight line segment set and the coordinates of the knife arm straight line segment set in the three-dimensional image according to the mapping relationship between the image to be identified and the three-dimensional image; a second output unit, used to output the three-dimensional coordinates as the positions of each contact and each knife arm of the target knife switch in the three-dimensional image.

9. An electronic device, characterized in that: The electronic device includes: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected via a line; The at least one processor calls the instructions in the memory to enable the electronic device to execute each step of the knife switch opening and closing state identification method according to any one of claims 1 to 5.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, each step of the method for identifying the switch opening and closing states as described in any one of claims 1 to 5 is implemented.

Images