Disconnector closed state detection method based on preset position camera

Abstract

The present invention discloses a method for detecting the closed state of an isolating switch based on a preset position camera, which obtains the preset position information of the camera, and the preset position information includes the type of knife switch monitored by the camera and the target area; performs feature matching and comparison between the monitoring image of the camera and the preset position image to determine whether the monitoring image of the camera deviates; if there is a deviation, the camera angle is adjusted based on the preset position image; otherwise, it is determined whether there is a target knife switch to be detected in the preset position image; if there is a target knife switch, a corresponding detection model is selected according to the type of the target knife switch; the monitoring image is input into the selected detection model, and the detection model gives the status result of the current knife switch according to the monitoring image; it is determined whether the degree of overlap between the status result of the current knife switch and the preset target knife switch position is greater than a set value; if so, the knife switch status detected by the model is output. The present invention does not require the addition of new sensors, and has a stronger generalization capability based on the deep learning method, can cover various weather and light changes, and can solve the problems existing in the prior art.

Description

Technical Field

[0001] The present invention relates to the technical field of power transformation monitoring, and in particular to a method for detecting the closed state of an isolating switch based on a preset position camera. Background Art

[0002] With the development of technology, power construction has developed rapidly. The opening and closing of disconnectors in substations are particularly important in power operation. In the extensive power grid, the manual inspection and operation of disconnectors is labor-intensive and the high-voltage environment is dangerous.

[0003] Existing methods for detecting the closed state of disconnectors mainly include image-based detection, a method for determining whether the disconnector closing operation is in place based on machine vision, and other sensor detection methods, including a method for detecting the disconnector opening and closing positions based on a posture sensor.

[0004] The first judgment method is based on machine vision. It trains a deep learning model to locate the position of the knife switch. The preliminary positioning of the knife switch is obtained based on the model detection results. Then, the knife switch image is contour analyzed in the local area of ​​the position to obtain the knife switch contour. In the located knife switch area, the knife switch image is processed into a grayscale image in sequence through traditional image processing methods. The image edge gradient is calculated, the gradient value is non-maximum suppressed, the image edge set is obtained, and the edge lines are extracted from the gradient image. The angle of the line is used to judge the closing state of the knife switch. The second is a disconnector opening and closing detection method based on a posture sensor. It mainly relies on the posture sensor to record the actual opening / closing angle of the disconnector during the opening / closing process, and calculate the angle change to judge the opening and closing state of the knife switch.

[0005] Existing technologies, such as those using gesture sensors, increase equipment operating costs and reduce resource utilization. Furthermore, connecting sensors to the switch poses certain risks. Existing approaches use image detection and deep learning models to determine the switch's location. The core function of determining the switch's closed state relies primarily on traditional image processing, utilizing image preprocessing, edge processing, line extraction, and angle calculation. However, in real-world scenarios, complex environments, often subject to significant interference from lighting and weather, pose limited generalization challenges for traditional image processing, making it difficult to adapt to detection tasks in complex environments. Summary of the Invention

[0006] The object of the present invention is to provide a method for detecting the closed state of an isolating switch based on a preset position camera.

[0007] The technical solution adopted in the present invention is:

[0008] The method for detecting the closed state of an isolating switch based on a preset position camera comprises the following steps:

[0009] Step 1: Obtain the preset position information of the camera, which includes the type of knife switch monitored by the camera and the target area;

[0010] Step 2: perform feature matching and comparison between the camera's monitoring image and the preset position image to determine whether the camera's monitoring image deviates; if so, adjust the camera angle based on the preset position image and execute step 2; otherwise, execute step 3;

[0011] Step 3, determine whether there is a target knife switch to be detected in the preset position image; if yes, select the corresponding detection model according to the type of the target knife switch and execute step 4; otherwise, stop the detection;

[0012] Step 4: Input the monitoring image into the selected detection model, and the detection model gives the current status result of the knife switch according to the monitoring image;

[0013] Step 5 determines whether the overlap between the current switch status and the preset target switch position is greater than a set value. If so, the switch status detected by the model is output. Otherwise, the confidence level of the selected detection model is lowered and step 4 is executed. The confidence level is a threshold that determines the detection result. Each detection result of the model has a confidence level. Detection results above the confidence level are output, while those below the confidence level are discarded. If there is no detection result that matches the target position, the threshold is lowered to expand the detection results.

[0014] Furthermore, the method for generating the preset position information in step 1 is: connecting the camera to the configuration tool provided by the system, and checking whether the knife switch appears completely in the monitoring screen; when the knife switch appears completely in the monitoring screen, capturing the preset position image, setting the camera preset position in the configuration tool and marking the knife switch category monitored by the camera, so as to generate the preset position and upload it to the system database through the configuration tool.

[0015] When the knife switch appears completely on the monitoring screen, it means that the preset position ensures that complete knife switch information can be obtained.

[0016] Furthermore, the status results of the knife switch in step 4 include the knife switch being closed, the knife switch being open, and the knife switch being not in the open or closed position.

[0017] Furthermore, in step 4, different types of disconnectors are tested separately (such as horizontal double opening, horizontal telescopic, vertical telescopic, etc.), and models of different types of switch are used for testing according to the pre-configured monitoring information of the camera.

[0018] Furthermore, the setting value in step 5 is 75%, that is, the selection standard is: the overlapping area between the yolov5 model detection result and the configuration tool frame is greater than 75%.

[0019] Furthermore, the detection model is the yolov5 model.

[0020] The present invention adopts the above technical solution and uses the monitoring equipment in the substation to collect real-time images of the disconnector for remote monitoring and analysis, which can greatly improve this problem. The present invention transmits real-time image materials through on-site monitoring video, performs visual image analysis through deep learning, detects the disconnection and closing status of the disconnector, and optimizes the detection results in combination with the camera preset position information. The main application areas are substation auxiliary detection, industrial automation and other aspects. The present invention does not require the addition of new sensors, and the deep learning-based method has stronger generalization capabilities, can cover various weather and light changes, and can solve the problems existing in the existing technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments;

[0022] Figure 1 Schematic diagram of the flow of the method for detecting the closed state of an isolating switch based on a preset position camera of the present invention;

[0023] Figure 2 This is a schematic diagram of a pre-positioned image capture according to the present invention;

[0024] Figure 3 Schematic diagram for detecting the closed state of the knife switch in the preset position information assistance. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions 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.

[0026] like Figure 1 As shown, the present invention discloses a method for detecting the closed state of an isolating switch based on a preset position camera, which includes the following steps:

[0027] Step 1: Obtain the preset position information of the camera, which includes the type of knife switch monitored by the camera and the target area;

[0028] Step 2: perform feature matching and comparison between the camera's monitoring image and the preset position image to determine whether the camera's monitoring image deviates; if so, adjust the camera angle based on the preset position image and execute step 2; otherwise, execute step 3;

[0029] Step 3, determine whether there is a target knife switch to be detected in the preset position image; if yes, select the corresponding detection model according to the type of the target knife switch and execute step 4; otherwise, stop the detection;

[0030] Step 4: Input the monitoring image into the selected detection model, and the detection model gives the current status result of the knife switch according to the monitoring image;

[0031] Step 5, such as Figure 3 As shown, the process determines whether the degree of overlap between the current switch state and the preset target switch position is greater than a set value. If so, the switch state detected by the model is output. Otherwise, the confidence level of the selected detection model is lowered and step 4 is executed. Confidence is a threshold that determines the detection result. Each detection result of the model has a confidence level. Results above the confidence level are output, while those below the confidence level are discarded. If no detection result matches the target position, the threshold is lowered to expand the detection results.

[0032] Further, if Figure 2 As shown, the method for generating the preset position information in step 1 is: connect the camera to the configuration tool provided by the system, check whether the knife switch appears completely in the monitoring screen; when the knife switch appears completely in the monitoring screen, capture the preset position image, set the camera preset position in the configuration tool and mark the knife switch category monitored by the camera to generate the preset position and upload it to the system database through the configuration tool.

[0033] When the knife switch appears completely on the monitoring screen, it means that the preset position ensures that complete knife switch information can be obtained.

[0034] Furthermore, in step 3, different types of disconnectors are tested separately (such as horizontal double opening, horizontal telescopic, vertical telescopic, etc.), and models of different types of switch are used for detection based on the pre-configured monitoring information of the camera. The present invention trains detection models with corresponding switch states for different types of switch.

[0035] Furthermore, the status results of the knife switch in step 4 include the knife switch being closed, the knife switch being open, and the knife switch being not in the open or closed position.

[0036] Furthermore, the setting value in step 5 is 75%, that is, the selection standard is: the overlapping area between the yolov5 model detection result and the configuration tool frame is greater than 75%.

[0037] Furthermore, the detection model is the yolov5 model.

[0038] The present invention adopts the above technical solution. Compared with the existing machine vision and posture sensor methods, the present invention has the following technical advantages: (1) It can cover different types of knife gates such as horizontal double opening, horizontal telescopic, and vertical telescopic; (2) It does not use traditional image processing methods to extract edges, lines, and calculate angles. Instead, it uses the Yolov5 model to directly detect the knife gate status, which has a stronger system generalization ability. (3) No new sensor equipment is added; (4) It uses a camera preset configuration system to adjust the camera monitoring angle so that each camera can be clearly and fully displayed in the monitoring image.

[0039] Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. In the absence of conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of the present application is not intended to limit the scope of the application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

Claims

1. A method for detecting the closed state of an isolating switch based on a preset position camera, characterized in that: It includes the following steps: Step 1: Obtain the preset position information of the camera, which includes the type of knife switch monitored by the camera and the target area; Step 2: perform feature matching and comparison between the camera's monitoring image and the preset position image to determine whether the camera's monitoring image deviates; if so, adjust the camera angle based on the preset position image and execute step 2; Otherwise, go to step 3; Step 3: determine whether there is a target knife switch to be detected in the preset position image; if yes, select the corresponding detection model according to the type of the target knife switch and execute step 4; Otherwise, stop testing; Step 4: Input the monitoring image into the selected detection model, and the detection model gives the current status result of the knife switch according to the monitoring image; Step 5: Determine whether the overlap between the current switch state result and the preset target switch position is greater than a set value; if so, output the switch state detected by the model; otherwise, reduce the confidence of the selected detection model and execute step 4; Among them, the method for generating the preset position information in step 1 is: connect the camera to the configuration tool provided by the system, check whether the knife switch appears completely in the monitoring screen; when the knife switch appears completely in the monitoring screen, capture the preset position image, set the camera preset position in the configuration tool and mark the knife switch category monitored by the camera to generate the preset position and upload it to the system database through the configuration tool.

2. The method for detecting the closed state of an isolating switch based on a preset position camera according to claim 1, characterized in that: In step 3, different types of disconnectors are tested separately, and models of different types of switch are used for testing based on the pre-configured monitoring information of the camera.

3. The method for detecting the closed state of an isolating switch based on a preset position camera according to claim 1, characterized in that: The status results of the knife switch in step 4 include the knife switch closed, the knife switch open, and the knife switch not being in place.

4. The method for detecting the closed state of an isolating switch based on a preset position camera according to claim 1, characterized in that: In step 5, set the value to 75%.

5. The method for detecting the closed state of an isolating switch based on a preset position camera according to claim 1, characterized in that: The detection model is the yolov5 model.

Images