Insulator automatic positioning method suitable for unmanned aerial vehicle zero value detection work

Abstract

The invention discloses an automatic insulator positioning method suitable for zero-value detection work of an unmanned aerial vehicle, and belongs to the field of image recognition positioning, and the method comprises the steps: positioning a detected target through an RTK positioning system, capturing an image of the detected target as a first input, outputting a first output with the coordinate information of the detected target, and outputting a second output with the coordinate information of the detected target; and extracting the offset of the measured target from the first output, and outputting a second output with the target type. Through combination of an image or visual technology and an RTK positioning system, centimeter-level and even millimeter-level positioning precision of the unmanned aerial vehicle on the insulator is realized, reliable identification and positioning of the insulator under different illumination conditions and complex backgrounds can be realized through the image or visual technology, the environmental adaptability of the unmanned aerial vehicle is improved, and the unmanned aerial vehicle can be used for positioning the insulator. The automatic image processing and recognition process reduces manual intervention, and improves the detection efficiency and accuracy.

Description

Technical Field

[0001] This invention relates to the field of image recognition and positioning, and specifically to an automatic insulator positioning method suitable for zero-value detection work of unmanned aerial vehicles. Background Technology

[0002] Insulators are indispensable key components in power systems, supporting conductors and preventing current leakage. However, with the continuous expansion of power grids and the increase in service life, insulators may experience aging, damage, or fouling, leading to a decline in insulation performance, i.e., the appearance of zero-value insulators. The presence of zero-value insulators can seriously affect the safe operation of power systems; therefore, regular zero-value testing of insulators is crucial. Traditional methods for zero-value insulator testing mainly rely on manual tower climbing, which is inefficient and carries high safety risks. Especially in high-altitude, strong electromagnetic field environments, manual testing is not only labor-intensive but also difficult to guarantee accuracy. In recent years, with the development of drone technology, using drones for insulator testing has become a new trend. Existing drone-based zero-value testing does not consider the issue of automatic insulator positioning, relying on manual drone operation for positioning, resulting in low testing efficiency and failing to meet the timeliness requirements of on-site work.

[0003] The comparison document (CN113219312A) relies on the UAV's own RTK positioning, but the position of the insulator needs to be manually observed or roughly estimated, which is easily affected by environmental interference (such as obstruction or excessive distance) and thus results in large positioning errors.

[0004] Prior art document (CN105160669B) discloses a method for detecting and locating insulator defects in images of power transmission lines inspected by unmanned aerial vehicles (UAVs). This method involves converting the image from RGB color space to HSI chromaticity-saturation-luminance space for image processing, and then using an ant colony algorithm based on particle swarm optimization parameters to locate the defective insulator. However, this method employs a large-model algorithm and does not consider the offset between the insulator and the conductor during actual measurement, resulting in low detection efficiency and failing to meet practical operating conditions. Summary of the Invention

[0005] The purpose of this invention is to achieve preliminary positioning of a UAV in a target area by using an RTK positioning system based on existing technology, guide the UAV to the vicinity of an insulator string, and use visual recognition technology to accurately position the insulator.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0007] An automatic insulator positioning method suitable for zero-value detection by unmanned aerial vehicles (UAVs) includes:

[0008] The target under test is located using an RTK positioning system.

[0009] Taking the captured image of the target as the first input, the output is a first output containing the coordinate information of the target.

[0010] Extract the offset of the target under test from the first output, and output a second output with the type of the target under test.

[0011] In the above technical solution, the video stream is used as the first input, and the first output has a set of several detection boxes. The center coordinate information of each detection box is calculated.

[0012] In the above technical solution, the target under test is determined to be within the detection frame. The center point of the target under test is used as a reference to obtain the relative offset of other targets and determine the offset type of the target under test.

[0013] In the above technical solution, the first output includes the coordinate information of the center point of the conductor end suspension point and the coordinate information of the center point of the pole end suspension point. The slope of the straight line containing the center point of the conductor end suspension point and the center point of the pole end suspension point is used as the first determination criterion to determine the target type of the second output.

[0014] In the above technical solution, the first determination criterion is:

[0015]

[0016] in: The determination coefficient is when It is determined to be a suspension string insulator when It was determined to be a tension insulator at that time. The coordinates of the center point of the pole end hanging point are: These are the coordinates of the center point of the conductor end suspension point.

[0017] This invention also discloses an automatic insulator positioning method suitable for zero-value detection work of unmanned aerial vehicles, comprising:

[0018] The target under test is located using an RTK positioning system.

[0019] Taking the captured image of the target as the first input, the output is a first output containing the coordinate information of the target.

[0020] Extract the offset of the measured target from the first output, and output a second output with the target type.

[0021] Using the offset as the second criterion, a third output is given, which is the correction amount for capturing the target image.

[0022] In the above technical solution, the target to be tested is determined to be within the detection frame. The center point of the target to be tested is used as a reference to obtain the relative offset of other targets. The relative offset is used to obtain the target image capture correction amount.

[0023] In the above technical solution, the second determination criterion includes: taking the center point of the captured image as a reference and the border of the product display image as a limit, determining the area within the image where the target point coordinates are located, and the correction amount within the production plane.

[0024] This invention also discloses an automatic insulator positioning method suitable for zero-value detection work of unmanned aerial vehicles, comprising:

[0025] The target under test is located using an RTK positioning system.

[0026] Taking the captured image of the target as the first input, the output is a first output containing the coordinate information of the target.

[0027] Extract the offset of the measured target from the first output, and output a second output with the target type.

[0028] Using the offset as the second criterion, a third output is generated, which is the correction amount for capturing the target image.

[0029] Feedback is provided via the third output to adjust the position of the image capture relative to the target being measured.

[0030] In the above technical solution, the proportion of the target in the first input image is determined, and control commands are generated according to the proportion to adjust the relative position of the UAV, thereby adjusting the position relative to the target when the image is captured.

[0031] This invention provides an automatic insulator positioning system suitable for zero-value detection work by unmanned aerial vehicles (UAVs), comprising:

[0032] It has at least one first module for image processing of a first input, wherein the first input is a video stream captured by a drone through a high-definition camera of the target under test, and the first module uses an improved YOLOv8 algorithm to perform multi-target detection on the video image and outputs a set of detection boxes;

[0033] It has at least one second module that determines the type of the target under test in the first output, and determines the determination criteria based on multiple suspension points related to the suspension of the target under test, and outputs the determination type as suspension string insulator or tension insulator.

[0034] It has at least one third module that corrects the offset of the measured target in the first output. This module is used to calculate the direction and amount of correction required based on the coordinates of the suspension point, thereby controlling the UAV to adjust its position relative to the measured target and complete the correction.

[0035] The principle of this invention is as follows: a drone takes pictures of the target under test, and based on the captured images, the element information in the images is extracted, the element information is judged and calculated, thereby generating control commands to control the drone to take pictures of the target under test in a loop, thereby obtaining accurate information about the target under test.

[0036] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0037] Improve positioning accuracy: By combining image or vision technology with RTK positioning systems, drones can achieve centimeter-level or even millimeter-level positioning accuracy for insulators.

[0038] Enhanced environmental adaptability: Image or vision technology can reliably identify and locate insulators under different lighting conditions and complex backgrounds, thereby improving the environmental adaptability of UAVs.

[0039] Improved detection efficiency: Automated image processing and recognition processes reduce manual intervention and improve detection efficiency and accuracy. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0041] Figure 1 This is a schematic diagram of the system framework of Embodiment 1;

[0042] Figure 2 This is a schematic diagram of the system framework of Embodiment 2;

[0043] Figure 3 This is a schematic diagram of the system framework of Embodiment 3;

[0044] Figure 4 This is a schematic diagram of the positioning process in an embodiment. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0046] The implementation of this embodiment is as follows: Figure 1 , Figure 2 , Figure 3 As shown, it includes:

[0047] The first module, used for processing video images, processes the video images and automatically identifies the anchor points, insulator strings, and conductors using the YOLOv8 algorithm. After identifying the anchor points, insulator strings, and conductors, it obtains their coordinates in the image.

[0048] The second module, which processes the output coordinate information, determines the type of the photographed target by defining the judgment coefficient and combining the coordinate values ​​of each suspension point, and outputs the type of the target as either a suspension string insulator or a tension insulator.

[0049] The third module is used to correct the offset of the measured target. It calculates the direction and amount of correction required based on the coordinates of the suspension point, and then uses the feedback to control the UAV to adjust its position relative to the measured target to complete the correction.

[0050] The control module is used to control the flight attitude of the UAV, adjust its position relative to the target by combining the feedback input of the third module, control the first and second modules to execute control commands, calculate and obtain the precise position and type of the target, and communicate with the host system through the communication module to ensure the real-time acquisition of measurement results.

[0051] In this embodiment, the first input is a video data stream captured in real time by a drone. , With a resolution of 3-channel RGB, the first module processes the video data stream and outputs a first output image with coordinate information.

[0052] In this embodiment, the image of the first output is used as the input of the second module. The second module is used to analyze and judge the coordinate information, identify the various offsets of the target in the image, and determine the type of the target by calculating the offset coordinates and use it as the second output.

[0053] In this embodiment, the first output image is used as input to the third module. The third module is used to correct the offset and feed the correction back to the control module. The control module controls the flight attitude of the UAV, adjusts the relative position of the image, and re-captures the image so that the target being measured is always in the center of the image in the final output image, ensuring accurate positioning.

[0054] Example 1

[0055] This embodiment combines Figure 1 , Figure 4 As shown

[0056] S1: The UAV is initially located in the target area using the RTK positioning system, and then guided to the vicinity of the insulator string.

[0057] As the drone flies towards the target insulator string, it uses the YOLOv8 algorithm to automatically identify the anchor points, insulator string, and conductor. After identifying the anchor points, insulator string, and conductor anchor points, it obtains their coordinates in the image and calculates their positional offset relative to the center point of the insulator string. This ensures that the entire insulator string is within the frame.

[0058] In this embodiment, target detection is first performed on the insulator string.

[0059] S2: Real-time video data streaming via drone cameras ( The first input is the coordinates of the center point of the insulator string (with a resolution of 3-channel RGB). The first output is the coordinates of the center point of the insulator string. .

[0060] In this embodiment, since the first input is a video data stream, the first output includes several consecutive image frames, each of which has a center point coordinate.

[0061] This embodiment uses an improved YOLOv8 algorithm to process all images from several data streams. Perform multi-object detection and output a set of boxes containing the detected objects. .

[0062] in, ,category {Tower end attachment point, insulator string, conductor end attachment point}, confidence threshold is generally considered This value can be adjusted according to the actual situation in practical applications.

[0063] S3: In this embodiment, for each detection box of the first output The coordinates of the center point are calculated as follows:

[0064]

[0065] If an insulator string frame is detected in these detection frames Then, based on its center point Using this as a baseline, calculate the relative offsets of other targets:

[0066] , , ( {Pole end hanging point, insulator string, conductor end hanging point}).

[0067] In this embodiment, based on the obtained first and second outputs, two types of insulator strings are given: suspension insulator strings (vertically installed) and tension insulator strings (horizontally installed), and an insulator string type determination coefficient is defined.

[0068]

[0069] when It was determined to be a suspension string insulator.

[0070] when It was determined to be a tension insulator at that time.

[0071] in: The determination coefficient, The horizontal axis position of the center point of the pole end hanging point. The vertical axis position of the center point of the pole end hanging point. The horizontal axis position is the center point of the conductor end suspension point. This refers to the vertical axis position of the center point of the conductor end suspension point.

[0072] like Figure 1 As shown, after the video stream data is processed by the first and second modules, it outputs insulator string information with coordinate information and classification type, realizing the location of the insulator string and indicating the type of the insulator string.

[0073] Example 2

[0074] This embodiment combines Figure 2 , Figure 4 As shown

[0075] S1: The UAV is initially located in the target area using the RTK positioning system, and then guided to the vicinity of the insulator string.

[0076] As the drone flies towards the target insulator string, it uses the YOLOv8 algorithm to automatically identify the anchor points, insulator string, and conductor. After identifying the anchor points, insulator string, and conductor anchor points, it obtains their coordinates in the image and calculates their positional offset relative to the center point of the insulator string. This ensures that the entire insulator string is within the frame.

[0077] In this embodiment, target detection is first performed on the insulator string.

[0078] S2: Real-time video data streaming via drone cameras ( The first input is the coordinates of the center point of the insulator string (with a resolution of 3-channel RGB). The first output is the coordinates of the center point of the insulator string. And the image overlay mark.

[0079] In this embodiment, since the first input is a video data stream, the first output includes several consecutive image frames, each of which has a center point coordinate.

[0080] This embodiment uses the YOLOv8 algorithm to process all images from several data streams. Perform multi-object detection and output a set of boxes containing the detected objects. .

[0081] in, ,category {Tower end attachment point, insulator string, conductor end attachment point}, confidence threshold is generally considered This value can be adjusted according to the actual situation in practical applications.

[0082] S3: In this embodiment, for each detection box of the first output The coordinates of the center point are calculated as follows:

[0083]

[0084] If an insulator string frame is detected in these detection frames Then, based on its center point Using this as a baseline, calculate the relative offsets of other targets:

[0085] , , ( {Pole end hanging point, insulator string, conductor end hanging point}).

[0086] S4: The insulator image in the second output image is centrally controlled. By judging the positional relationship between the tower end attachment point, the conductor end attachment point, and the insulator center position in the image captured by the drone, the drone position is continuously corrected to ensure that the insulator to be inspected is completely within the image. The specific implementation process is as follows:

[0087] In this embodiment, the image captured by the drone is defined as a rectangular area, with its center point coordinates as follows: The screen width is The height is Define the boundaries of the screen as:

[0088] Left boundary: Right boundary:

[0089] Upper boundary: Lower boundary:

[0090] The center coordinates of the tower end attachment point obtained in the first output Coordinates of the center of the conductor end hanging point and the center coordinates of the insulator string Make the following judgment:

[0091] If there exists a point whose coordinates satisfy: or or or This indicates that the point is not in the image. For example, if This indicates that the pole end attachment point is outside the left side of the image.

[0092] Determine the drone's correction direction based on the position of points not in the frame. Let the drone's position correction amount be... The revised rules are as follows:

[0093] Horizontal correction

[0094] like ,but ,in This is the horizontal correction factor. At this point, the drone should move to the right to include the pole end attachment point in the frame.

[0095] like ,but At this point, the drone should move to the left to include the pole end attachment point in the frame.

[0096] Similarly, for the center coordinates of the conductor end hanging point... and the center coordinates of the insulator string Calculate the corresponding horizontal correction amount respectively. .

[0097] Vertical correction

[0098] like ,but ,in This is the vertical correction factor. At this point, the drone should move upwards to include the attachment point at the end of the pole in the frame.

[0099] like ,but At this point, the drone should move to the left to include the pole end attachment point in the frame.

[0100] Similarly, for the center coordinates of the conductor end hanging point... and the center coordinates of the insulator string Calculate the corresponding horizontal correction amount respectively. .

[0101] Taking into account the correction requirements at the three points, the largest correction amount in each direction is taken as the final correction amount, that is:

[0102]

[0103] Where: sign (direction to be corrected) determines the positive or negative value based on which side of the screen the point is outside. For example, if the point is outside the left side of the screen, the horizontal correction is positive (making the drone move to the right), and if the point is outside the right side of the screen, the horizontal correction is negative (making the drone move to the left); if the point is outside the top of the screen, the vertical correction is positive (making the drone move downwards), and if the point is outside the bottom of the screen, the vertical correction is negative (making the drone move upwards).

[0104] Example 3

[0105] Based on Example 2, such as Figure 3 , Figure 4 As shown, the correction amount for the drone is obtained by feeding back the third output. This triggers a course adjustment for the drone. The control module controls the drone's flight attitude based on the correction amount, adjusting its shooting position relative to the target.

[0106] Define the historical position correction amount for the drone:

[0107] Define control coefficient: proportional coefficient Speed ​​feedback coefficient

[0108] Based on the correction amount of the measured target Calculate the PD control quantity:

[0109]

[0110] in, This indicates the time interval between two consecutive executions of a control command.

[0111] For the calculated Constraints are applied sequentially:

[0112] Step distance constraint:

[0113]

[0114] Where clip is the clipping function.

[0115] Velocity constraint: converting displacement into velocity After being limited, the displacement is restored.

[0116]

[0117] The current correction amount and speed are recorded, and the constrained displacement is sent to the UAV's underlying controller. The underlying system performs the following operations: adjusts the motor speed or thrust vector to achieve spatial displacement triggering hardware-level safety monitoring.

[0118] After each control cycle is completed, the system waits for new correction values ​​and mode flags, then re-executes the process to form a closed-loop control, continuously adjusting the trajectory until the insulator is completely centered on the screen. Then, the screen is captured again, and the above judgment and correction process is repeated until the tower end suspension point, conductor end suspension point, and the center point of the insulator string are all within the screen.

[0119] In the above three embodiments, the optimal insulator string detection and acquisition position is determined by the actual proportion of the insulator string in the image.

[0120] In the above embodiments, an RTK positioning system is used to achieve preliminary positioning of the UAV in the target area, guiding the UAV to the vicinity of the insulator string for coarse positioning. Based on RTK positioning, visual recognition technology is combined for precise positioning. The UAV adjusts its flight attitude and position according to the visual recognition results until a high-definition camera captures a clear image of the insulator string, and the specific position of the insulator is confirmed by image processing algorithms, achieving precise positioning.

[0121] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An automatic insulator positioning method suitable for zero-value detection by unmanned aerial vehicles (UAVs), characterized in that: The target under test is located using an RTK positioning system. Taking the captured image of the target as the first input, the output is a first output containing the coordinate information of the target. Extract the offset of the target under test from the first output, and output a second output with the type of the target under test.

2. The automatic insulator positioning method for zero-value detection of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that... Using the video stream as the first input, the output is a first output containing a set of detection boxes, and the center coordinate information of each detection box is calculated.

3. The automatic insulator positioning method for zero-value detection of unmanned aerial vehicles (UAVs) according to claim 2, characterized in that... Once the target is confirmed to be within the detection frame, the relative offsets of other targets are obtained using the center point of the target as a reference, and the type of offset of the target is determined.

4. An automatic insulator positioning method for zero-value detection work of unmanned aerial vehicles (UAVs) according to any one of claims 1-3, characterized in that... The first output includes the coordinates of the center point of the conductor end suspension point and the center point of the pole end suspension point. The slope of the straight line containing the center points of the conductor end suspension point and the pole end suspension point is used as the first criterion to determine the target type of the second output.

5. The automatic insulator positioning method for zero-value detection of unmanned aerial vehicles (UAVs) according to claim 2, characterized in that... The first criterion is: ， in: The determination coefficient is when It is determined to be a suspension string insulator when It was determined to be a tension insulator at that time. The coordinates of the center point of the pole end hanging point are: These are the coordinates of the center point of the conductor end suspension point.

6. An automatic insulator positioning method suitable for zero-value detection by unmanned aerial vehicles (UAVs), characterized in that: The target under test is located using an RTK positioning system. Taking the captured image of the target as the first input, the output is a first output containing the coordinate information of the target. Extract the offset of the measured target from the first output, and output a second output with the target type. Using the offset as the second criterion, a third output is given, which is the correction amount for capturing the target image.

7. The automatic insulator positioning method for zero-value detection of unmanned aerial vehicles (UAVs) according to claim 6, characterized in that... Using the video stream as the first input, the output is a first output containing a set of detection boxes, and the center coordinate information of each detection box is calculated.

8. The automatic insulator positioning method for zero-value detection of unmanned aerial vehicles (UAVs) according to claim 7, characterized in that... Once the target is identified within the detection frame, the relative offsets of other targets are obtained using the center point of the target as a reference. The image capture correction amount of the target is then obtained using these relative offsets.

9. An automatic insulator positioning method for zero-value detection of unmanned aerial vehicles (UAVs) according to claim 7, characterized in that... The second determination criteria include: taking the center point of the captured image as a reference and the border of the product display image as a limit, determining the area within the image where the target point coordinates are located, and the correction amount within the production plane.

10. An automatic insulator positioning method suitable for zero-value detection by unmanned aerial vehicles (UAVs), characterized in that: The target under test is located using an RTK positioning system. Taking the captured image of the target as the first input, the output is a first output containing the coordinate information of the target. Extract the offset of the measured target from the first output, and output a second output with the target type. Using the offset as the second criterion, a third output is generated, which is the correction amount for capturing the target image. Feedback is provided via the third output to adjust the position of the image relative to the target being measured during image capture.

Images