Millimeter wave radar power transmission line feature detection method and system for unmanned aerial vehicle power grid inspection scene

Abstract

The invention belongs to the field of unmanned aerial vehicle power transmission line fault detection, and discloses a millimeter-wave radar power transmission line feature detection method and system for an unmanned aerial vehicle power grid inspection scene, and the method comprises the steps: installing a millimeter-wave radar at the belly position of an unmanned aerial vehicle, enabling the view field plane of the radar to be perpendicular to a to-be-detected power transmission line, and drawing a distance angle heat map; collecting a large amount of data, and making a data set for deep learning model training; and training a Yolov8 model, and detecting distance and angle information of a power transmission line target in the heat map. According to the millimeter wave radar power transmission line feature detection and processing method for the unmanned aerial vehicle power transmission line fault detection scene, stable power transmission line feature detection can be realized in all weathers and all scenes.

Description

Technical Field

[0001] The present invention belongs to, but is not limited to, the technical field of UAV power line fault detection, and particularly relates to a millimeter-wave radar power line feature detection method and system for UAV power grid inspection scenarios. Background Art

[0002] Using UAVs for power grid inspection can not only significantly improve the efficiency of inspection operations, but also effectively avoid the safety risks faced by inspection personnel, and has high engineering practical value. At present, the most commonly used method in UAV inspection is monocular vision detection. The pilot operates the UAV to fly along the power line, uses a single camera to collect images and perform fault detection. However, traditional vision-based power line detection methods cannot obtain the distance information of the target power line, and existing depth sensor-based detection methods are also difficult to achieve stable detection of power lines under actual inspection conditions. This not only limits the ability of the UAV to safely approach a specific cable, but also hinders the acquisition of close-range power line images, thus affecting the development of fault detection work.

[0003] The FMCW millimeter-wave radar emits a continuous frequency-modulated signal, receives the reflected signal and mixes it with the transmitted signal to generate an intermediate-frequency signal. By analyzing the frequency and phase of the intermediate-frequency signal, the distance, speed and angle of the target can be accurately measured, realizing high-precision detection and tracking of the target. Millimeter-wave sensors usually use the automotive radar frequency band of 24 - 81 GHz. At these wavelengths, the reflections of most non-metallic materials such as plastics and plants will become weaker, while the reflections of metals will be stronger, which is a useful characteristic when detecting metal cables. Due to the influence of weather, lightning, and birds, the fault location of power line cables is usually above the power line. Therefore, the UAV needs to take pictures of the power line from top to bottom to better obtain the operating state of the cable. The key problem in applying millimeter-wave radar to UAV power line detection is that when the radar scans from top to bottom, since the power line and the ground are relatively stationary, and there are clutter reflections from objects such as trees and stones on the ground, conventional radar target detection algorithms such as CFAR cannot effectively separate the power line. Therefore, it is of great significance to propose a method for detecting power lines from top to bottom using millimeter-wave radar.

[0004] Therefore, aiming at the problem that the current UAV-borne power line detection method cannot effectively obtain the depth information of the power line under actual inspection conditions, the present invention proposes a millimeter-wave radar power line feature detection method and system for UAV power grid inspection scenarios.

[0005] In order to obtain the distance between the aircraft and the power line, existing detection technologies mainly use multi-sensors to assist in depth detection, including binocular cameras, lidar, infrared TOF cameras, and millimeter-wave radars. However, the current detection technologies have the following problems:

[0006] (1) A binocular camera uses two cameras separated by a known distance to find corresponding points in the images of the two cameras and calculate the pixel position difference of the points in the two images, thereby calculating the distance to these points. Since it is based on an optical camera, it is easily affected by lighting conditions, such as mountain fog, and the power transmission line is prone to specular reflection of sunlight. Therefore, stable distance detection cannot be achieved in the actual inspection environment;

[0007] (2) A lidar uses a laser beam to scan the surrounding environment and determines the distance and azimuth to the target by receiving the reflected laser. The advantages of this sensor are high resolution and high precision, so it is widely used in the field of environmental mapping. The disadvantages are large data volume, high cost, which pose high requirements for the data processing ability of the UAV on-board computer, and it is also relatively sensitive to atmospheric environmental factors;

[0008] (3) An infrared TOF camera is an imaging device that uses the time-of-flight principle to measure distance. This camera emits infrared light pulses and measures the time when the reflected light returns, thereby calculating the distance of the object. However, this sensor will produce complex multipath effects when encountering complex tree and rock backgrounds in the mountains, and strong sunlight will also interfere with the infrared light emitted by the sensor.

[0009] (4) A millimeter-wave radar emits electromagnetic waves in the millimeter-wave band and receives the reflected signals, and measures the distance, speed, and angle information of the target by analyzing the frequency and phase of the reflected signals. Although this sensor is basically not affected by the atmospheric environment and lighting, in the actual inspection environment, it is easily affected by complex background echoes, resulting in difficulty in stably detecting power transmission lines by traditional radar target detection methods.

[0010] In view of the above analysis, the technical problems that need to be urgently solved in the existing technology are:

[0011] The above methods based on multiple depth sensors have poor detection effects in the actual inspection environment and cannot achieve stable power transmission line detection in areas with poor atmospheric environment, poor lighting conditions, and complex backgrounds. Summary of the Invention

[0012] In view of the problems existing in the prior art, the present invention provides a millimeter-wave radar power transmission line feature detection method and system for UAV power grid inspection scenarios.

[0013] The present invention is implemented as follows. A millimeter-wave radar power transmission line feature detection method for a UAV power grid inspection scenario, characterized in that the millimeter-wave radar power transmission line feature detection method for a UAV power grid inspection scenario specifically includes:

[0014] S1: Install the millimeter-wave radar at the belly position of the drone, make the radar's field-of-view plane perpendicular to the power transmission line to be detected, and draw a range-angle heat map.

[0015] S2: Collect a large amount of data and create a dataset for deep learning model training.

[0016] S3: Train the Yolov8 model to detect the distance and angle information of the power transmission line target in the heat map.

[0017] Furthermore, in S1, use the drone to carry the millimeter-wave radar AWR1642. The millimeter-wave radar is installed at the belly position of the drone, and the radar's detection plane is perpendicular to the nose direction of the drone. The airborne computer controls the millimeter-wave radar to transmit a frequency-modulated continuous wave signal and receive the echo signal. After mixing and ADC sampling, the data is sent to the airborne computer, and then distance FFT and angle FFT are respectively performed on the data. After coordinate system conversion and interpolation, the radar's range-azimuth heat map is drawn.

[0018] Furthermore, in S2, create a dataset based on the millimeter-wave radar range-angle heat map. Control the drone to fly at multiple different relative positions with respect to the power transmission line, collect sufficient data, perform smooth interpolation on the data matrix after distance FFT and angle FFT using the spline16 algorithm, and perform color mapping with jet; preprocess the obtained data, screen out valid data, and increase the diversity of the data through data augmentation; after annotating the target area in the data, divide the data into a training set, a validation set, and a test set.

[0019] Furthermore, in S3, use a deep learning-based method for target detection. The detection model selects Yolov8. The model is trained using the training set in the dataset. After training, input the data to be detected, output the detection result including the target detection bounding box, and take the coordinates of the maximum value of the echo signal within the box as the detection result.

[0020] Another object of the present invention is to provide a millimeter-wave radar power transmission line feature detection system for the drone power grid inspection scenario. The system specifically includes:

[0021] An inspection platform for installing the millimeter-wave radar, processing radar data, and drawing a range-angle heat map.

[0022] A dataset construction module for creating a dataset for deep learning model training.

[0023] A target detection module for training the model and generating detection results.

[0024] Combined with the above technical solutions and the solved technical problems, the advantages and positive effects of the technical solution to be protected by the present invention are:

[0025] First, the present invention proposes a method for detecting the depth information of transmission lines in the scenario of UAV inspection of transmission lines. A millimeter-wave radar is installed at the belly position of the UAV, enabling the UAV to inspect at a safe distance above the transmission line, which is more conducive to detecting faults in the upper part of the transmission line.

[0026] The present invention designs the data acquisition path of the inspection UAV, acquires a sufficient amount of feature maps for the transmission line fault detection scenario, normalizes the radar-based feature maps to obtain a uniform distance-angle spectrum, and constructs a deep learning offline dataset through data augmentation, data annotation, and segmentation.

[0027] The present invention proposes a transmission line detection algorithm based on millimeter-wave radar feature maps. By using a deep learning-based method to detect transmission line targets in the feature maps generated by the radar, it can achieve detection performance higher than that of ordinary CFAR detection algorithms in cases with complex background clutter interference such as mountainous areas.

[0028] The present invention proposes a method for simultaneously detecting multiple transmission lines using millimeter-wave radar. The inspection UAV flies along the transmission line, and multiple transmission lines are simultaneously covered within the scanning plane of the radar. The detection output not only includes the presence or absence of the target but also provides the distance and angle information of each transmission line, making the detection results more comprehensive and detailed.

[0029] Second, the expected benefits and commercial values after the transformation of the technical solution of the present invention are as follows: The application prospect of using UAVs for transmission line detection is very broad. First, using UAVs for transmission line inspection can easily cross complex terrains such as high mountains and canyons, avoiding the potential safety hazards to personnel caused by manual climbing of poles for inspection. Second, applying millimeter-wave radar to UAV inspection can effectively detect the depth information of transmission lines, making the UAV safer when approaching the transmission line. In addition, since the radar is not affected by light and atmospheric environment, it can be used as a more robust detection method to improve the problems encountered in optical inspection when the optical inspection scheme works unsatisfactorily. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flowchart of the millimeter-wave radar transmission line feature detection method for the UAV power grid inspection scenario provided by an embodiment of the present invention;

[0031] Figure 2 is a module diagram of the millimeter-wave radar transmission line feature detection system for the UAV power grid inspection scenario provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0033] As Figure 1 shown, an embodiment of the present invention provides a millimeter-wave radar transmission line feature detection method for the UAV power grid inspection scenario. The method specifically includes:

[0034] S1: Install the millimeter-wave radar at the belly position of the UAV, make the radar's field-of-view plane perpendicular to the transmission line to be detected, and draw a distance-angle heat map;

[0035] S2: Collect a large amount of data and make a data set for deep learning model training;

[0036] S3: Train the Yolov8 model to detect the distance and angle information of the transmission line target in the heat map.

[0037] The method of the present invention first installs the millimeter-wave radar at the belly position of the UAV to ensure that the radar's field-of-view plane is perpendicular to the transmission line to be detected. The millimeter-wave radar emits electromagnetic waves and receives reflected signals, collects the distance information and angle information of the target, and generates a high-precision distance-angle heat map. The heat map will reflect the characteristic distribution of the transmission line and surrounding targets, forming visual detection data to ensure that the system can stably obtain target data during the UAV flight.

[0038] During the UAV inspection process, the millimeter-wave radar continuously collects a large amount of distance-angle heat map data. In order to train the deep learning detection model, these data need to be processed and labeled to construct a high-quality data set. Specifically, it includes target labeling and classification of transmission lines, poles and interference objects in the heat map to form labeled samples. To improve the generalization performance of the detection model, data augmentation techniques, such as noise processing, angle change, occlusion simulation, etc., can be used during the data set construction process to expand the training sample size and ensure that the model can adapt to various complex scenarios.

[0039] Input the constructed data set into the deep learning object detection network Yolov8 for training. Yolov8 is a high-performance real-time object detection algorithm with the characteristics of fast detection and high accuracy. Through multiple rounds of training and parameter optimization, the model can learn the characteristic rules of the transmission line in the heat map and form an accurate object detection ability. During the actual detection process, the Yolov8 model analyzes the heat map generated by the millimeter-wave radar, automatically detects and outputs the target position of the transmission line, that is, the distance and angle information, providing data support for the positioning and inspection results of the transmission line.

[0040] The working principle of the whole method is completed through the coordination of four steps: "millimeter-wave radar data acquisition - dataset construction - deep learning model training - target detection". In the scenario of drone inspection, the distance-angle heat map drawn by the millimeter-wave radar is detected in real time through the Yolov8 model, and the system can quickly extract the distance and angle information of the transmission line target and output accurate positioning results. Compared with the traditional optical detection method, this method has stronger anti-interference ability, is applicable to complex environments such as rain, fog, and low light, and significantly improves the automation, intelligence, and efficiency of power grid inspection.

[0041] In S1, a drone is used to carry the millimeter-wave radar AWR1642. The millimeter-wave radar is installed at the belly position of the drone, and the detection plane of the radar is perpendicular to the nose direction of the drone. The airborne computer controls the millimeter-wave radar to transmit a frequency-modulated continuous wave signal and receive the echo signal. After mixing and ADC sampling, the signal is sent to the airborne computer, and then distance FFT and angle FFT are respectively performed on the data. After coordinate system conversion and interpolation, the distance-azimuth heat map of the radar is drawn.

[0042] During the inspection process, the drone flies along the line above the transmission line. At this time, the transmission line is perpendicular to the detection plane of the radar. The transmission line reflects the electromagnetic wave signal emitted by the radar. Due to the different distances between the transmission line and multiple receiving antennas, the phase change of the signals received by different receiving antennas is caused. After distance FFT and angle FFT, a higher echo signal will be generated in the area where the transmission line is located, and the actual distance and angle information can be obtained through coordinate system conversion and coordinate axis mapping.

[0043] In S2, to achieve object detection based on deep learning, a dataset based on the millimeter-wave radar distance-angle heat map is made. The drone is controlled to fly at multiple different relative positions with respect to the transmission line to collect sufficient data. The spline16 algorithm is used for smoothing interpolation on the data matrix after distance FFT and angle FFT, and jet is used for color mapping. The obtained data is preprocessed to filter out valid data, and data augmentation is used to increase the diversity of the data to improve the generalization ability of the model. Finally, after annotating the target area in the data, the data is divided into a training set, a validation set, and a test set.

[0044] In S3, since the data collected by the millimeter-wave radar is affected by ground clutter, a deep learning-based method is used for object detection, and the detection model is selected as Yolov8. The model is trained using the training set in the dataset. After training, the data to be detected is input, and the detection result containing the object detection bounding box is output. The coordinates of the maximum value of the echo signal within the box are taken as the detection result.

[0045] As Figure 2As shown in the figure, a millimeter-wave radar transmission line feature detection system for the UAV power grid inspection scenario provided by an embodiment of the present invention specifically includes:

[0046] An inspection platform for installing a millimeter-wave radar, processing radar data, and drawing a distance-angle heat map;

[0047] A dataset construction module for making a dataset for deep learning model training;

[0048] A target detection module for training a model and generating detection results.

[0049] The inspection platform of the present invention is used to carry a millimeter-wave radar and serves as the core module for data acquisition. This platform can be integrated into the UAV system to achieve high-altitude inspection operations on the power grid transmission line. The millimeter-wave radar collects high-precision distance and angle information of the transmission line by transmitting and receiving electromagnetic waves, generating raw radar data. The platform processes the collected data in real time and draws a distance-angle heat map through an algorithm to visually present the characteristic distribution of the transmission line and surrounding targets, providing data support for subsequent detection.

[0050] The dataset construction module is responsible for making a dataset for deep learning model training to ensure that the system has good target detection capabilities. This module constructs a high-quality training sample library by annotating, classifying, and processing the raw data collected by the millimeter-wave radar. Specifically, it includes feature marking and classification of targets such as transmission lines, poles, and obstacles to generate a labeled dataset. In addition, to improve the generalization ability of the model, the dataset construction module can also perform data augmentation processing, including noise addition, perspective change, occlusion simulation, etc., which is applicable to various complex inspection environments.

[0051] The target detection module is the part that realizes the core function in the system and is mainly responsible for training and applying the deep learning model. The module trains on the constructed dataset to generate a target detection model with high-precision detection capabilities. During the inspection process, the target detection module analyzes the distance-angle heat map drawn by the millimeter-wave radar in real time, identifies and locates the transmission line and related target features, and outputs the detection results. Through real-time feedback and optimization of the detection results, the system can effectively distinguish the transmission line, surrounding objects, and potential obstacles to ensure the accuracy and stability of the detection.

[0052] In the embodiments of the present invention, through the collaborative work of the inspection platform, the dataset construction module, and the target detection module, the millimeter-wave radar transmission line feature detection for the UAV power grid inspection scenario is realized. The system can quickly and efficiently obtain the target features of the transmission line in a complex environment and generate intuitive detection results. Compared with the traditional optical inspection method, the system has the ability to work all-weather and is applicable to harsh weather conditions such as rain, fog, and low light. In addition, through the training of the deep learning model and the fusion of millimeter-wave radar data, the system significantly improves the accuracy and automation level of transmission line detection, promoting the development of the intelligentization of power grid inspection.

[0053] I. Specific application fields or related products of the present invention

[0054] 1. Application fields:

[0055] UAV power grid inspection: Suitable for UAVs equipped with millimeter-wave radars to conduct inspections on transmission lines to ensure the safety of transmission lines and the surrounding environment.

[0056] Power system maintenance: Through accurate identification of transmission line features, automated detection and inspection are realized, improving the maintenance efficiency of transmission lines.

[0057] Power grid intelligent management: Provide basic data support for the smart grid and promote the digital and intelligent transformation of the power grid.

[0058] Application of millimeter-wave radar: Promote the application of millimeter-wave radars in fields such as power, communication, and rail transit, especially target detection and tracking in complex environments.

[0059] 2. Related products:

[0060] UAV power inspection system: Includes UAV platform, millimeter-wave radar module, data processing platform, etc.

[0061] Smart grid inspection equipment: Equipment integrating millimeter-wave radar, deep learning target detection model, data processing, and monitoring terminal.

[0062] Millimeter-wave radar detection system: An independent module or integrated into UAVs and inspection robots to detect features such as transmission lines, poles, and surrounding obstacles.

[0063] II. Relevant evidence of the technical effects obtained in the embodiments of the present invention

[0064] 1. Embodiment of technical effects:

[0065] High-precision target detection: Through the combination of millimeter-wave radar and deep learning model, the features of the transmission line (such as cable position, shape, etc.) can be accurately detected, reducing the false detection rate and missed detection rate.

[0066] Adapt to complex environments: Millimeter-wave radar has all-weather detection capabilities, can penetrate harsh weather conditions such as rain and fog, and ensure the stability and accuracy of inspection data.

[0067] Intelligent data processing: By drawing distance-angle heat maps and training deep learning models, automatically analyze and identify the characteristics of transmission lines to improve detection efficiency.

[0068] Automated and efficient inspection: The drone-mounted system can automatically execute tasks, replace manual inspections, improve operation efficiency, reduce costs and human risks.

[0069] 2. Evidence support points:

[0070] Experimental data and results:

[0071] Compare the detection accuracy, applicable environmental range, and processing speed of drone-mounted millimeter-wave radar with traditional optical inspections.

[0072] Specific examples:

[0073] Improved detection accuracy: At a certain detection distance, the detection accuracy of millimeter-wave radar reaches over 90%;

[0074] Strong anti-interference ability: In rainy, foggy weather or low-light conditions, the detection results maintain high stability;

[0075] Automated model output: Comparing the detection results generated by deep learning algorithms with manually labeled data, the error is controlled within 5%.

[0076] Actual scenario application examples:

[0077] Conduct drone inspections in a certain power grid area and successfully detect defects or foreign objects on high-voltage transmission lines;

[0078] The drone can automatically complete inspection tasks under the set flight path, provide real-time feedback of radar images and detection results, and achieve automatic early warning of grid abnormal information.

[0079] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated designed hardware. Those of ordinary skill in the art can understand that the above-mentioned devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code is provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuits of programmable hardware devices such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, or can be implemented by a combination of the above hardware circuits and software such as firmware.

[0080] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A millimeter wave radar transmission line feature detection method for UAV power grid inspection scenarios, characterized in that: The method specifically includes: S1: Install the millimeter-wave radar on the belly of the drone, make the radar's field of view perpendicular to the power line to be detected, and draw a distance angle heat map; S2: Collect a large amount of data and create a dataset for deep learning model training; S3: Train the Yolov8 model to detect the distance and angle information of the power line targets in the heat map.

2. The millimeter wave radar transmission line feature detection method for UAV power grid inspection scenarios as claimed in claim 1 is characterized in that: The S1 uses a UAV equipped with a millimeter-wave radar AWR1642. The millimeter-wave radar is installed on the belly of the UAV. The detection plane of the radar is perpendicular to the direction of the UAV nose. The millimeter-wave radar is controlled by an onboard computer to transmit a frequency-modulated continuous wave signal and receive an echo signal. After mixing and ADC sampling, the signal is sent to the onboard computer. The data is then subjected to distance FFT and angle FFT respectively. After coordinate system conversion, the radar's distance and azimuth heat map is obtained by interpolation.

3. The millimeter wave radar transmission line feature detection method for UAV power grid inspection scenarios as claimed in claim 1 is characterized in that: The S2 prepares a data set based on the millimeter-wave radar distance angle heat map, controls the drone to fly at multiple different relative positions to the power line, collects sufficient data, uses the spline16 algorithm to smoothly interpolate the data matrix after distance FFT and angle FFT, and uses jet for color mapping; preprocesses the obtained data, filters valid data, and increases data diversity through data enhancement; after marking the target area in the data, divides the data into a training set, a validation set, and a test set.

4. The millimeter wave radar transmission line feature detection method for UAV power grid inspection scenarios as claimed in claim 1 is characterized in that: The S3 uses a deep learning-based method to perform target detection, and the detection model selects Yolov8. The model is trained using a training set in the data set. After the training, the data to be detected is input, and the detection result containing the target detection boundary box is output. The coordinates of the maximum value of the echo signal in the box are taken as the detection result.

5. A millimeter wave radar transmission line feature detection system for drone power grid inspection scenarios as claimed in claim 14, characterized in that: The system specifically includes: A millimeter wave radar module is installed on the belly of the drone, and the millimeter wave radar is used to transmit a frequency modulated continuous wave signal and receive an echo signal reflected by a transmission line; The data acquisition module is used to collect the echo signal of the millimeter wave radar, and perform distance FFT and angle FFT processing on the echo signal to generate a distance azimuth heat map; A deep learning target detection module is used to train the Yolov8 model and detect the location of the transmission line target based on the generated range-azimuth heat map; The data processing module is used to interpolate, smooth, color map and mark the target area of ​​the collected data, and generate a training data set for the deep learning model.

6. The system of claim 51, wherein: The millimeter-wave radar module is an AWR1642 millimeter-wave radar, which is installed on the belly of the drone. The millimeter-wave radar detection plane is perpendicular to the direction of the drone's nose. The millimeter-wave radar is controlled by an onboard computer for data acquisition and signal processing.

7. The system according to claim 5, characterized in that The data acquisition module comprises: A data preprocessing unit, used for smoothing interpolation and color mapping of the distance FFT and angle FFT data matrices generated by the millimeter wave radar; Data screening unit, used to screen valid data and increase the diversity of data sets through data enhancement; The data segmentation unit is used to segment the data into training sets, validation sets, and test sets for training deep learning models.

8. The system according to claim 5, characterized in that The deep learning target detection module includes: Yolov8 model training unit, used to train the Yolov8 target detection model using the training set; Yolov8 detection unit, which is used to input the detection data into the trained Yolov8 model and output the target detection results; The target positioning unit is used to calculate the coordinates of the maximum value of the echo signal based on the target detection boundary box to determine the distance and angle information of the transmission line target.

Images