A positioning installation method and device of a bird repelling device, a terminal equipment and a computer readable storage medium

Abstract

The application discloses a positioning installation method and device of a bird repelling device, terminal equipment and a computer readable storage medium, and belongs to the technical field of power line operation and maintenance and unmanned aerial vehicle intelligent control. The method comprises the following steps: acquiring GNSS, IMU and image data of an unmanned aerial vehicle; performing electromagnetic noise filtering on the GNSS data; calculating a signal-to-noise ratio and a feature matching degree to dynamically adjust a fusion weight, wherein the GNSS weight is positively correlated with the signal-to-noise ratio, the IMU weight is negatively correlated with the signal-to-noise ratio, and the visual weight is positively correlated with the matching degree; and outputting accurate coordinates and postures based on the adjusted weight by using an extended Kalman filter, and controlling the unmanned aerial vehicle to hover and the mechanical arm posture to complete installation. Through electromagnetic noise suppression and a multi-source weight adaptive compensation strategy, the application effectively solves the problems of low installation precision and high operation risk caused by positioning drift and mechanical shaking in a high-voltage line strong electromagnetic interference and signal fluctuation environment.

Description

Technical Field

[0001] This invention relates to the fields of power line operation and maintenance and intelligent control technology for unmanned aerial vehicles, and in particular to a method, device, terminal equipment and computer-readable storage medium for positioning and installing a bird deterrent device. Background Technology

[0002] At present, bird control of power lines mainly relies on installing bird deterrent devices on the poles. However, traditional installation methods often involve manual climbing of the poles or live-line work vehicles, which have problems such as high operational risks, high labor intensity and severe terrain limitations.

[0003] However, existing UAV positioning technologies face severe challenges when operating in environments with high-voltage power lines and complex terrain. On the one hand, strong electromagnetic interference exists around power lines, causing GNSS signals to drift or lose lock, making single satellite positioning insufficient for centimeter-level installation accuracy. On the other hand, existing multi-source fusion positioning technologies (such as GPS + vision + inertial navigation) often employ fixed weighting strategies or simple threshold switching logic, lacking a mechanism to dynamically adjust the trust level of each sensor based on the intensity of electromagnetic interference and visual environment characteristics. This "hard switching" or "fixed weighting" approach often fails to effectively eliminate noise when the signal is in an intermediate state of "interference but not completely lost," leading to jumps in positioning results and causing UAV attitude instability, making it difficult to ensure the robotic arm's accurate capture of tiny installation points. Summary of the Invention

[0004] This invention provides a positioning and installation method for a bird deterrent device, which can solve the problems of low positioning accuracy and low installation success rate caused by the lack of targeted noise suppression and multi-source fusion weight adaptive adjustment mechanism in the context of strong electromagnetic interference and signal fluctuation.

[0005] An embodiment of the present invention provides a method for positioning and installing a bird deterrent device, comprising: Acquire GNSS positioning data, IMU inertial navigation data, and environmental image data collected by visual sensors at the current flight location of the UAV; The GNSS positioning data is subjected to electromagnetic noise filtering to obtain the first positioning data; The signal-to-noise ratio of the first positioning data is calculated in real time, and the feature matching degree between the environmental image data and the preset tower feature information is calculated. Based on the signal-to-noise ratio and the feature matching degree, a first weight corresponding to the first positioning data, a second weight corresponding to the IMU inertial navigation data, and a third weight corresponding to the environmental image data are determined respectively; wherein, the first weight is positively correlated with the signal-to-noise ratio, the third weight is positively correlated with the feature matching degree, and the second weight is negatively correlated with the signal-to-noise ratio; Based on the first weight, the second weight, and the third weight, the first positioning data, the IMU inertial navigation data, and the environmental image data are fused and calculated to output the three-dimensional coordinates and attitude information of the UAV. Based on the three-dimensional coordinates and attitude information, the drone is controlled to hover and the attitude of the drone's robotic arm is adjusted to install the bird deterrent device at the preset point on the target tower.

[0006] Further, the step of performing electromagnetic noise filtering on the GNSS positioning data to obtain the first positioning data specifically includes: determining the target stopband frequency range based on the power line's characteristic frequency and harmonic frequency range; constructing a band-stop filter for the target stopband frequency range; using the band-stop filter to perform frequency domain filtering on the GNSS positioning data, suppressing signal components within the target stopband frequency range, and outputting the first positioning data after filtering out interference.

[0007] Furthermore, calculating the feature matching degree between the environmental image data and the preset tower feature information specifically includes: Construct a preset feature library containing a subset of feature point descriptors for key components of the target tower; extract local feature points within the current field of view based on the environmental image data; match the local feature points with the subset of feature point descriptors in the preset feature library and calculate the number of successfully matched feature points; map the number of successfully matched feature points to the feature matching degree.

[0008] Further, determining the first weight corresponding to the first positioning data, the second weight corresponding to the IMU inertial navigation data, and the third weight corresponding to the environmental image data based on the signal-to-noise ratio and the feature matching degree specifically includes: Obtain the first baseline weight value of the first weight, the second baseline weight value of the second weight, and the third baseline weight value of the third weight; wherein the sum of the first baseline weight value, the second baseline weight value, and the third baseline weight value is 1; When the signal-to-noise ratio is greater than a preset first signal-to-noise ratio threshold, the first weight is increased from the first baseline weight value to the first dominant weight value, the second weight is decreased from the second baseline weight value to the first auxiliary weight value, and the third weight is decreased from the third baseline weight value to the second auxiliary weight value; wherein, the first auxiliary weight value and the second auxiliary weight value are both less than the first dominant weight value, and the sum of the first dominant weight value, the first auxiliary weight value, and the second auxiliary weight value is 1; When the signal-to-noise ratio (SNR) is between a preset first SNR threshold and a preset second SNR threshold, it is determined whether the feature matching degree meets the preset reliability condition. If yes, the third weight is increased from the third baseline weight value to the visual compensation weight value, the first weight is decreased from the first baseline weight value to the GNSS suppression weight value, and the second weight remains unchanged at the second baseline weight value. If no, the first weight, the second weight, and the third weight remain unchanged at their corresponding baseline weight values. The sum of the visual compensation weight value, the GNSS suppression weight value, and the second baseline weight value is 1. When the signal-to-noise ratio is less than the second signal-to-noise ratio threshold, the first weight is reduced from the first baseline weight value to zero, the second weight is increased from the second baseline weight value to the second dominant weight value, and the third weight is adjusted from the third baseline weight value to the visual correction weight value according to the feature matching degree; wherein, the visual correction weight value is positively correlated with the feature matching degree, and the sum of the visual correction weight value and the second dominant weight value is 1.

[0009] Further, the step of fusing the first positioning data, the IMU inertial navigation data, and the environmental image data according to the first weight, the second weight, and the third weight to output the three-dimensional coordinates and attitude information of the UAV specifically includes: constructing a state equation containing the UAV's position and velocity states, and an observation equation containing the first positioning data, the IMU inertial navigation data, and the environmental image data; dynamically adjusting the observation noise covariance matrix parameters of the corresponding observations in the extended Kalman filter algorithm used in the fusion calculation according to the first weight, the second weight, and the third weight; and using the adjusted observation noise covariance matrix parameters to perform state prediction and update, and output the three-dimensional coordinates and attitude information.

[0010] Furthermore, the robotic arm is a gravity-oriented flexible connection structure; the step of controlling the drone to hover and adjusting the robotic arm attitude based on the three-dimensional coordinates and attitude information specifically includes: controlling the flight position of the drone based on the three-dimensional coordinates, so that the drone hovers above a preset point on the target tower; monitoring the instantaneous sway amplitude of the drone based on the three-dimensional coordinates, attitude information, and IMU inertial navigation data; when the sway amplitude exceeds a preset safety range causing the robotic arm to swing, controlling the motor speed of the drone to perform reverse attitude compensation, and actively suppressing the swing of the robotic arm.

[0011] Furthermore, the end of the robotic arm is equipped with a variable-force electromagnetic adsorption mechanism. The installation of the bird-repelling device at a preset point on the target pole specifically includes: adjusting and controlling the posture of the robotic arm to push the bird-repelling device to the preset point and detecting the contact pressure in real time; when the contact pressure meets the installation conditions, controlling the electromagnetic adsorption mechanism to release the bird-repelling device so that the bird-repelling device is installed at the preset point on the target pole; wherein, the driving current of the electromagnetic adsorption mechanism is determined according to the load type information of the bird-repelling device.

[0012] Another embodiment of the present invention provides a positioning and installation device for a bird deterrent device, comprising: a data acquisition module, a filtering processing module, a quality assessment module, a weight adjustment module, a fusion calculation module, and an installation control module; The data acquisition module is used to acquire GNSS positioning data, IMU inertial navigation data, and environmental image data collected by the visual sensor at the current flight location of the UAV. A filtering module is used to perform electromagnetic noise filtering on the GNSS positioning data to obtain first positioning data; The quality assessment module is used to calculate the signal-to-noise ratio of the first positioning data in real time, and to calculate the feature matching degree between the environmental image data and the preset target feature information; The weight adjustment module is used to determine, based on the signal-to-noise ratio and the feature matching degree, a first weight corresponding to the first positioning data, a second weight corresponding to the IMU inertial navigation data, and a third weight corresponding to the environmental image data; wherein, the first weight is positively correlated with the signal-to-noise ratio, the third weight is positively correlated with the feature matching degree, and the second weight is negatively correlated with the signal-to-noise ratio; The fusion computing module is used to perform fusion computing on the first positioning data, the IMU inertial navigation data and the environmental image data based on the first weight, the second weight and the third weight, using the extended Kalman filter algorithm, and output the three-dimensional coordinates and attitude information of the UAV. The installation control module is used to control the drone to hover and adjust the attitude of the drone's robotic arm based on the three-dimensional coordinates and attitude information, so as to install the bird deterrent device at the preset point of the target pole.

[0013] Another embodiment of the present invention provides a terminal device, including: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements the steps of the positioning and installation method of the bird deterrent device of the present invention.

[0014] Another embodiment of the present invention provides a computer-readable storage medium item, including: a stored computer program, wherein, when the computer program is running, it controls the device where the computer-readable storage medium is located to execute the positioning and installation method of the bird deterrent device of the present invention.

[0015] The embodiments of the present invention have the following beneficial effects: This invention provides a method for positioning and installing a bird deterrent device. By performing electromagnetic noise filtering on the acquired GNSS positioning data, electromagnetic interference noise unique to power line environments can be filtered out at the source. This eliminates false signal interference when the UAV acquires the first positioning data, providing a cleaner and more reliable position reference for subsequent fusion. Based on this, this application establishes a dynamic weight determination mechanism based on signal quality and feature quality. By setting a logical relationship where the first weight is positively correlated with the signal-to-noise ratio (SNR), the second weight is negatively correlated with the SNR, and the third weight is positively correlated with the feature matching degree, the sensor positioning is achieved. The adaptive allocation of reliability means that when the GNSS signal-to-noise ratio decreases, its weight is automatically reduced and the weight of IMU inertial navigation data is increased to compensate, or when the feature matching degree is high, the weight of environmental image data is increased to enhance positioning. Thus, when the quality of a single source decreases or fluctuates, it can automatically rely on other high-quality sources to maintain the continuity and stability of positioning. Finally, by directly using the three-dimensional coordinates and attitude information output by the fusion calculation to control the drone's hovering and adjust the attitude of the robotic arm, it is ensured that the actuator can correct the deviation in real time based on the fused comprehensive attitude information, and the bird deterrent device is accurately installed at the preset point on the target tower. Attached Figure Description

[0016] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1This is a schematic flowchart of a positioning and installation method for a bird deterrent device provided in an embodiment of the present invention.

[0018] Figure 2 This is a schematic diagram of the structure of a bird deterrent device positioning and installation method provided in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0021] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0022] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0023] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0024] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0025] In the description of the embodiments of this application, unless otherwise expressly specified and limited, technical terms such as "installation," "connection," "joining," and "fixing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0026] To address the problems of low positioning accuracy and low installation success rate in existing technologies under strong electromagnetic interference and signal fluctuation environments due to the lack of targeted noise suppression and multi-source fusion weight adaptive adjustment mechanisms, an embodiment of the present invention provides a positioning and installation method for a bird deterrent device, comprising: Step S1: Acquire GNSS positioning data, IMU inertial navigation data, and environmental image data collected by the visual sensor at the current flight location of the UAV; In a preferred embodiment, in response to the complex electromagnetic environment and varied terrain features encountered in power line inspection and installation operations, a multi-dimensional perception network is constructed to simultaneously collect the absolute spatial position, relative attitude changes, and texture features of the external environment of the UAV.

[0027] The GNSS positioning data is acquired using a high-precision dual-mode satellite receiver mounted on the top of the UAV. Considering the potential for insufficient satellite acquisition by a single satellite system in mountainous or canyon terrain, this embodiment preferably employs a dual-mode receiver module supporting both BDS-3 and GPS. Specifically, the receiver is equipped with a high-sensitivity anti-interference antenna with a signal gain set to ≥25dB ​​and excellent axial ratio characteristics for receiving low-elevation satellite signals. Furthermore, to address the electromagnetic radiation environment unique to power lines, an IP67-rated electromagnetic shielding enclosure is used to physically isolate the direct coupling of external electric fields to the radio frequency circuitry. At the data output level, the GNSS receiver outputs raw messages (such as the NMEA-0183 protocol format) containing longitude, latitude, altitude, ground speed, heading angle, and positioning accuracy in real time at a high update rate of 10Hz. The collected GNSS dataset is denoted as: In the formula, These represent the drone's current longitude and latitude, respectively. This represents the current ground height of the drone, that is, its vertical height relative to the WGS-84 reference ellipsoid. These represent the eastward velocity component, northward velocity component, and skyward velocity component of the UAV in the Northeast Earth (NED) navigation coordinate system, respectively. This represents the position covariance matrix calculated from the satellite geometric distribution factor (DOP) (reflecting positional confidence). An absolute timestamp representing satellite time synchronization.

[0028] The inertial navigation data is acquired in real time by a six-axis MEMS (Micro-Electro-Mechanical Systems) inertial measurement unit mounted on the UAV's core control board. Because the UAV requires extremely high attitude stability when hovering to install bird deterrent devices, this embodiment rigorously selected the hardware specifications of the IMU: a high-precision industrial-grade chip with gyroscope zero-bias stability ≤0.1∘ / h (degrees / hour) and accelerometer zero-bias ≤10μg was chosen. The IMU includes a three-axis gyroscope and a three-axis accelerometer, used to measure the three-axis angular velocity and three-axis specific force of the UAV's body coordinate system relative to inertial space, respectively. This raw data reflects the UAV's dynamic response in a very short time, especially for high-frequency swaying caused by sudden gusts of wind or robotic arm movements. The IMU operates at a high-frequency sampling rate of 100Hz to 200Hz, and the acquired inertial navigation dataset is denoted as: In the formula, These represent the instantaneous angular rates of the UAV body along its body coordinate system (following the right-hand rule) along the X-axis (roll axis), Y-axis (pitch axis), and Z-axis (yaw axis), respectively. These represent the specific forces felt by the drone body along the three axes of its body coordinate system. The data sampling timestamp, representing the clock generated inside the inertial measurement unit, is usually the monotonically increasing counter value after the microcontroller is powered on. It should be noted that before the IMU data enters the subsequent algorithm, it also needs to undergo installation error angle compensation and lever effect compensation to unify the measurement center to the center of mass of the UAV.

[0029] The environmental image data is acquired using an industrial-grade binocular vision sensor mounted below the nose of the drone. To identify the outline of power towers at long distances (e.g., 100 meters away) and to clearly see the texture of crossarm bolts at close distances (e.g., within 5 meters), this embodiment uses a high-definition camera with a resolution of up to 4K and a frame rate of 30fps. The vision sensor employs global shutter exposure, which, compared to a rolling shutter, effectively avoids the rolling shutter effect caused by high-speed movement or vibration of the drone, ensuring that the geometric features of the image remain undistorted. Furthermore, the baseline distance of the binocular camera is precisely calibrated, enabling direct recovery of environmental depth information through the parallax principle. The acquired visual dataset includes left and right eye images, denoted as: In the formula, These represent the original image matrices of the left and right eyes captured in the k-th frame, respectively. The intrinsic parameter matrix representing the camera (including focal length, principal point coordinates, and distortion coefficients). A timestamp representing the moment the image was exposed; specifically, for power line backgrounds (usually the sky or complex mountains), the camera integrates an automatic exposure algorithm that focuses on metering the tower area in the center of the image to prevent the tower details from being underexposed and turning black due to an overly bright sky.

[0030] It is worth noting that, considering the significant differences in data sampling rates among multiple sensors (10Hz for GNSS, 200Hz for IMU, and 30Hz for vision) and the different time bases of each sensor (UTC time for GNSS, and relative time for IMU and camera data after system startup), this method strictly implements hard synchronization and soft alignment during the acquisition process. Specifically, the PPS (pulse per second) signal output from the GNSS receiver is used as a hardware trigger source. The FPGA or microcontroller simultaneously triggers the shutters of the IMU and camera using this pulse signal, forcibly unifying the time base of each sensor. In the data storage queue, the high-frequency time axis of the IMU is used as the main axis, and an interpolation algorithm is used to align the GNSS data and vision data to the most recent IMU sampling time, ensuring that the constructed fused data packet... The same physical moment describes the same motion state of the UAV. In addition, it is necessary to pre-calibrate the extrinsic parameter matrices (i.e., rotation matrices and translation vectors) between each sensor, and to uniformly transform the GNSS antenna phase center and camera optical center to the body coordinate system where the IMU measurement center is located, so as to eliminate the calculation errors caused by spatial position differences.

[0031] By synchronously acquiring multi-source heterogeneous data and strictly unifying spatiotemporal benchmarks, a multi-dimensional perception space covering macroscopic geographic coordinates, microscopic attitude dynamics, and intuitive environmental textures was constructed. This data acquisition method breaks through the vulnerability of traditional power operation drones relying solely on single GPS data, achieving deep information complementarity: GNSS data provides absolute geofencing constraints, ensuring the drone does not fly out of the work area; IMU data provides high-frequency attitude feedback, capable of sensitively capturing minute swaying that is imperceptible to the naked eye, providing millisecond-level input for subsequent pendulum oscillation control; visual data endows the drone with the ability to "identify the work object," enabling it to distinguish specific components of the power tower (such as crossarms and bolts) from background interference (such as tree branches). By integrating these three data sources, which are completely different in frequency domain characteristics (low frequency / high frequency), spatial attributes (absolute / relative), and information dimensions (numerical / image), the risk of single-point failure caused by electromagnetic interference from high-voltage lines can be effectively addressed. This lays a solid and highly redundant data foundation for subsequent dynamic weight allocation in Kalman filtering, achieving centimeter-level high-precision positioning and installation control.

[0032] Step S2: Perform electromagnetic noise filtering on the GNSS positioning data to obtain the first positioning data; In a preferred embodiment, the step of performing electromagnetic noise filtering on the GNSS positioning data to obtain first positioning data specifically includes: determining a target stopband frequency range based on the power line's characteristic frequency and harmonic frequency range; constructing a band-stop filter for the target stopband frequency range; and using the band-stop filter to perform frequency domain filtering on the GNSS positioning data to suppress signal components within the target stopband frequency range, and outputting the first positioning data after interference filtering.

[0033] Preferably, the target stopband frequency range is determined through the following steps: First, a reference power frequency is defined based on the standard power frequency of the power system. Based on the principle of electromagnetic induction, the first N harmonics with the most concentrated interference energy are determined (for example, N=40, covering up to 2000Hz), thus constructing a target stopband frequency set composed of multiple discrete frequency points. : Where f represents a continuous frequency variable in the frequency domain (in Hz), used to define the boundary of the stopband; n is the harmonic order. The set is a single-sided stopband bandwidth (e.g., ±2Hz) to cover small fluctuations in the grid frequency; this set clearly defines the range of noise spectrum that needs to be removed, avoiding phase delay or amplitude attenuation of real moving signals (such as low-frequency components generated by drone maneuvers) caused by broadband low-pass filtering.

[0034] Preferably, a band-stop filter for the target stopband frequency range is constructed through the following steps: In order to effectively filter out the above-mentioned specific frequency noise while maximizing the retention of the true dynamic information in the GNSS positioning data, this embodiment preferably adopts a cascaded IIR (Infinite Impulse Response) notch filter bank; Specifically, for the set each interference center frequency in where is the data sampling rate, a second-order IIR notch filter is designed, and the mathematical expression of its transfer function H(z) in the Z domain is: In the formula, and respectively represent the unit delay operator, r is the pole radius, and its value range is 0 < r < 1 (for example, r = 0.95); the parameter r directly determines the quality factor of the notch filter, that is, the closer r is to 1, the narrower the notch bandwidth and the smaller the impact on adjacent true signals.

[0035] The final band-stop filter model is cascaded by all N notch filters corresponding to harmonics, that is: This cascaded structure can form a comb-shaped amplitude-frequency response characteristic with multiple deep depression points, and accurately filter out power frequency and its harmonic interference.

[0036] Preferably, the first positioning data is obtained through the following steps: First, a time-domain difference equation is used to perform real-time recursive filtering on the collected GNSS raw sequence (including longitude, latitude, altitude, and velocity components). For the input sequence x[k] (i.e., the original GNSS data), the output sequence y[k] of the nth-order notch filter is calculated as follows: After N-stage cascaded processing, the output data is the first positioning data.

[0037] It should be particularly noted that in order to eliminate the influence of the inherent phase lag of the IIR filter on real-time control, in this embodiment, when the computing power permits, a forward-backward zero-phase filtering strategy is adopted, or a phase lead compensation link is introduced in the real-time system to ensure that the output first positioning data is strictly synchronized with the true physical movement on the time axis.

[0038] In this embodiment, the filtering process targeting specific electromagnetic spectrum characteristics described above preserves the high-frequency real signals of the UAV's rapid maneuvers (such as displacement when encountering a sudden gust of wind), while directionally eliminating periodic spurious noise generated by electromagnetic field coupling from high-voltage lines. The resulting first positioning data has a significantly improved signal-to-noise ratio, smooths out the spikes in the position curve without introducing additional time delay, and provides a clean sample for subsequent calculation of accurate signal-to-noise ratio indicators. This provides a high-confidence GNSS reference input for subsequent multi-source fusion algorithms.

[0039] Step S3: Calculate the signal-to-noise ratio of the first positioning data in real time, and calculate the feature matching degree between the environmental image data and the preset tower feature information; In a preferred embodiment, calculating the feature matching degree between the environmental image data and the preset tower feature information specifically includes: Construct a preset feature library containing a subset of feature point descriptors for key components of the target tower; extract local feature points within the current field of view based on the environmental image data; match the local feature points with the subset of feature point descriptors in the preset feature library and calculate the number of successfully matched feature points; map the number of successfully matched feature points to the feature matching degree.

[0040] Preferably, the signal-to-noise ratio of the first positioning data is calculated as follows: Although the GNSS receiver directly outputs the carrier-to-noise ratio (C / However, in high-voltage line environments, signal strength alone is insufficient to fully reflect the reliability of positioning. Therefore, the signal-to-noise ratio defined in this embodiment is a composite weighted index. Specifically, the satellite geometric precision factor (PDOP) and the number of visible satellites are extracted from the first positioning data after filtering in step S2. Theoretical maximum carrier-to-noise ratio constant ( ), number of saturated satellites in the system ( ) and average carrier-to-noise ratio ( The integrated signal-to-noise ratio is calculated using a normalized weighted formula. : in, , , As a weighting coefficient, this index can more comprehensively reflect the confidence level of GNSS positioning solutions when electromagnetic interference causes partial frequency band loss of lock or increased multipath effects; the higher the overall signal-to-noise ratio, the higher the confidence level.

[0041] Preferably, constructing a pre-defined feature library containing a subset of feature point descriptions of key components of the target tower specifically includes the following steps: In the offline phase before the operation, high-resolution sample images of the towers to be operated on (such as 10kV cement towers, angle steel towers, steel pipe towers, etc.) are collected from multiple angles. Feature extraction is performed on sample images using either the Scale Invariant Feature Transform (SIFT) algorithm or the Speed-Up Robust Feature Transform (SURF) algorithm. Specifically, an Extrema in scale space are detected by constructing a Difference-of-Gaussian Pyramid (DoG), and the gradient direction histogram in the neighborhood of each Extrema is calculated to generate a 128-dimensional feature vector, i.e., a feature point descriptor, which is invariant to rotation and scale. These descriptors and their corresponding physical spatial relative positions are stored in a database to form a pre-defined feature library. M represents the total number of feature points in the library; the library specifically covers key components with rich textures and fixed geometric structures, such as crossarm bolts, insulator string connections, and tower ladder edges.

[0042] Preferably, the step of extracting local feature points within the current field of view based on the environmental image data is as follows: During the real-time flight of the drone, each frame of environmental image data collected... The image is converted to grayscale and then processed using the same feature extraction algorithm (such as SIFT) used when building the feature library. First, the image is downsampled to construct an image pyramid, improving computational efficiency. Second, keypoints are detected at each layer of the pyramid, and low-contrast points and edge response points are removed to ensure the robustness of the extracted feature points. Finally, the principal direction of each keypoint is calculated, and a real-time local feature descriptor subset is generated based on the principal direction. , where K is the number of feature points extracted in the current field of view; this process can effectively overcome the image differences caused by changes in distance (scale scaling) and viewpoint (rotation) during the flight of the UAV.

[0043] Preferably, the local feature points are matched with the feature point descriptor subset in the preset feature library through the following steps, and the number of successfully matched feature points is calculated: The nearest neighbor search algorithm is used to calculate the real-time descriptor. Feature point descriptors in the feature library Euclidean distance between them: To improve matching speed, this embodiment preferably uses a kd-tree or FLANN (Fast Library for Approximate Nearest Neighbors) algorithm for fast indexing. After obtaining initial matching pairs, strict mismatch removal is necessary: ​​First, Lowe's Ratio Test is performed, meaning that a match is retained only if the ratio of the nearest neighbor distance to the second nearest neighbor distance is less than a preset threshold (e.g., 0.7), thus eliminating fuzzy features; Second, the RANSAC (Random Sample Consensus) algorithm is used to calculate the homography matrix based on epipolar geometric constraints, removing outliers that do not conform to the geometric transformation rules; After these two steps of filtering, the number of remaining interior points represents the number of successfully matched feature points. .

[0044] Preferably, the specific steps for mapping the number of successfully matched feature points to the feature matching degree are as follows: To transform the number of discrete feature points into a normalized weight adjustment factor, this step preferably uses a sigmoid function or a piecewise linear function for mapping; firstly, a preset locking threshold is set. (For example, the 15 points preset in this scheme), this threshold represents the minimum number of feature points required for the visual positioning algorithm to calculate a stable pose; the feature matching degree The specific calculation formula is as follows: ; in, The number of feature points that were successfully matched. Ideal number of matches (e.g., 50 points); Feature matching degree The value range is [0,1]. This piecewise function specifies that when the number of matching points is less than the preset locking threshold... When the matching degree is set to zero, inferior visual data is eliminated, which directly reflects the current visual system's recognition degree and confidence level of the target.

[0045] In this embodiment, by using high-dimensional feature matching of the SIFT algorithm and RANSAC geometric verification, it is possible to effectively distinguish real towers from trees or background noise that look similar, avoiding misidentification by traditional visual algorithms in complex backgrounds. The remaining quality of the GNSS signal after frequency domain filtering is further quantified by calculating the comprehensive signal-to-noise ratio. Furthermore, the precise quantification of the signal-to-noise ratio and feature matching degree index provides an effective mathematical basis for the dynamic weight allocation in the subsequent step S4, ensuring that the multi-source fusion algorithm can flexibly allocate different weights to different sensors according to the actual situation, thereby achieving the optimal positioning strategy in extreme environments.

[0046] Step S4: Based on the signal-to-noise ratio and the feature matching degree, determine the first weight corresponding to the first positioning data, the second weight corresponding to the IMU inertial navigation data, and the third weight corresponding to the environmental image data; wherein, the first weight is positively correlated with the signal-to-noise ratio, the third weight is positively correlated with the feature matching degree, and the second weight is negatively correlated with the signal-to-noise ratio; In a preferred embodiment, determining a first weight corresponding to the first positioning data, a second weight corresponding to the IMU inertial navigation data, and a third weight corresponding to the environmental image data based on the signal-to-noise ratio and the feature matching degree specifically includes: Obtain the first baseline weight value of the first weight, the second baseline weight value of the second weight, and the third baseline weight value of the third weight; wherein the sum of the first baseline weight value, the second baseline weight value, and the third baseline weight value is 1; When the signal-to-noise ratio is greater than a preset first signal-to-noise ratio threshold, the first weight is increased from the first baseline weight value to the first dominant weight value, the second weight is decreased from the second baseline weight value to the first auxiliary weight value, and the third weight is decreased from the third baseline weight value to the second auxiliary weight value; wherein, the first auxiliary weight value and the second auxiliary weight value are both less than the first dominant weight value, and the sum of the first dominant weight value, the first auxiliary weight value, and the second auxiliary weight value is 1; When the signal-to-noise ratio (SNR) is between a preset first SNR threshold and a preset second SNR threshold, it is determined whether the feature matching degree meets the preset reliability condition. If yes, the third weight is increased from the third baseline weight value to the visual compensation weight value, the first weight is decreased from the first baseline weight value to the GNSS suppression weight value, and the second weight remains unchanged at the second baseline weight value. If no, the first weight, the second weight, and the third weight remain unchanged at their corresponding baseline weight values. The sum of the visual compensation weight value, the GNSS suppression weight value, and the second baseline weight value is 1. When the signal-to-noise ratio is less than the second signal-to-noise ratio threshold, the first weight is reduced from the first baseline weight value to zero, the second weight is increased from the second baseline weight value to the second dominant weight value, and the third weight is adjusted from the third baseline weight value to the visual correction weight value according to the feature matching degree; wherein, the visual correction weight value is positively correlated with the feature matching degree, and the sum of the visual correction weight value and the second dominant weight value is 1.

[0047] Preferably, the baseline parameters for each weight are first obtained: The system reads the preset first baseline weight value. Second benchmark weight value and the third benchmark weight value This set of baseline values ​​represents the default trust allocation of the system in standard or standby states, and satisfies the normalization condition: + + =1 (for example, default) =0.4, =0.3, =0.3); Preferably, when the signal-to-noise ratio is greater than a preset first signal-to-noise ratio threshold, the specific weight adjustment logic is as follows: First, set the first signal-to-noise ratio threshold. (For example, 35dB) indicates that the GNSS signal quality is excellent.

[0048] The first weight is increased from the first benchmark weight value to the first dominant weight value. (For example, set to 0.7); The second weight is lowered from the second baseline weight value to the first auxiliary weight value. (For example, set to 0.15); The third weight is lowered from the third baseline weight value to the second auxiliary weight value. (For example, set to 0.15), and check the constraints. + + =1; By executing the weight adjustment logic in this step, high-confidence satellite positioning data can be used as an absolute position reference, with only IMU and visual data used for auxiliary smoothing, ensuring that the drone approaches the target tower with maximum efficiency in open areas.

[0049] Preferably, when the signal-to-noise ratio is between a preset first signal-to-noise ratio threshold and a preset second signal-to-noise ratio threshold, it is determined to be a signal occlusion state. In this case, the specific weight adjustment steps are as follows: Set the second signal-to-noise ratio threshold (For example, 25dB), this range corresponds to the region of enhanced electromagnetic interference; At this point, the GNSS suppression weight value is calculated based on the first benchmark weight value. This value typically decreases linearly as the signal-to-noise ratio decreases: Simultaneously, the feature matching degree α is detected to determine whether the confidence condition (e.g., α ≥ 0.6) is met: If the following conditions are met (visual recognition is available): Adjust the first weight to the GNSS suppression weight value calculated above. =0.4; Keep the second weight as the second baseline weight value. It remains unchanged (e.g., kept at 0.15), serving as the inertial base of the system; The third weight is increased to the visual compensation weight value. According to the normalization constraints + + =1, the system automatically calculates the visual compensation weight value: =1− - =0.45, thus enabling the use of visual increments to accurately fill the accuracy gaps lost by GNSS due to interference.

[0050] If the condition is not met (visual recognition is unavailable): then keep the weights unchanged as the baseline weight values ​​and wait for the signal to recover.

[0051] Preferably, when the signal-to-noise ratio is less than the second signal-to-noise ratio threshold, the specific weight adjustment strategy is as follows: At this point, the system determines that the GNSS data is no longer reliable (SNR < 25dB), and accordingly forces the first weight to be reduced to zero. ); Calculate the visual correction weight value based on the feature matching degree α (value range 0~1). This value is positively correlated with the matching degree, for example: =k·α (where k is an adjustment coefficient, such as 0.4 to 0.6, for example, taking k=0.4, α=0.85, at this time...) =0.34).

[0052] Increase the second weight to the second dominant weight value. .

[0053] Based on normalization constraints + =1, calculated as follows; =1− =0.66, At this point, the system is restructured into an IMU-dominated + visually corrected architecture, with IMU weights ( ) serves as the primary continuity guarantee (dominant), while visual weight ( The magnitude of the IMU drift correction is directly dependent on the degree of feature point recognition. The higher the degree of recognition, the stronger the correction of IMU drift. Conversely, it relies more on short-term IMU calculations.

[0054] In this embodiment, an active planning mechanism based on the confidence of multi-source sensors is established through the above-mentioned dynamic weight allocation strategy based on multi-dimensional features: the system generates a set of real-time, normalized confidence weight vectors according to the electromagnetic interference level (signal-to-noise ratio) and visual feature richness (matching degree) of the environment. This vector is no longer a static empirical value, but a dynamic parameter that strictly follows the complementary logical chain of GNSS suppression-visual compensation-IMU baseline, thereby providing key prior knowledge for the filter in the subsequent step S5. This allows the filter to obtain high-confidence sensor data at the current moment before performing mathematical calculations, thereby avoiding the divergence of calculations caused by blind fusion and laying the decision foundation for achieving robust positioning in complex environments.

[0055] Step S5: Based on the first weight, the second weight, and the third weight, perform fusion calculation on the first positioning data, the IMU inertial navigation data, and the environmental image data, and output the three-dimensional coordinates and attitude information of the UAV; In a preferred embodiment, the step of fusing the first positioning data, the IMU inertial navigation data, and the environmental image data according to the first weight, the second weight, and the third weight to output the three-dimensional coordinates and attitude information of the UAV includes: constructing a state equation containing the UAV's position and velocity states, and an observation equation containing the first positioning data, the IMU inertial navigation data, and the environmental image data; dynamically adjusting the observation noise covariance matrix parameters of the corresponding observations in the extended Kalman filter algorithm used in the fusion calculation according to the first weight, the second weight, and the third weight; and using the adjusted observation noise covariance matrix parameters to perform state prediction and update, and output the three-dimensional coordinates and attitude information. Preferably, the specific steps for constructing the state equation including the UAV's position and velocity states, and the observation equation including the first positioning data, the IMU inertial navigation data, and the environmental image data are as follows: First, a state equation containing the UAV's position and velocity states is constructed, as well as an observation equation containing the first positioning data, the IMU inertial navigation data, and the environmental image data; Define the system state vector of the UAV Its dimensions are designed to be 15 (or higher) to fully describe the kinematic features: ; in, The coordinates are three-dimensional (latitude, longitude, and altitude). It is a three-dimensional velocity vector. For three-dimensional attitude angles (Euler angles or quaternions). These are the zero-bias errors of the accelerometer and gyroscope, respectively, and a zero-bias term is introduced to estimate and compensate for the cumulative drift of the IMU online. Next, based on the high-frequency sampling characteristics of the IMU, a discretized state equation (time update model) is constructed: in, The control input is directly taken from the acceleration and angular velocity measurements of the IMU; f(·) is a nonlinear state transition function based on the inertial navigation mechanics arrangement; The process noise follows a Gaussian distribution; Then, construct the multi-source observation equations for measuring the updated model: in, is the comprehensive observation vector at the current moment, which includes the first positioning data from GNSS, the relative pose data from the visual sensor, etc.; h(·) is the observation function, which is used to map the state space to the observation space; To observe noise; It should be noted that the observation vector The structure is as follows: in, The first positioning data after filtering out electromagnetic interference; The relative displacement corresponding to the visual solution; The corresponding virtual observations are constructed using zero-rate correction or non-integrity constraints; Preferably, the specific calculation process for dynamically adjusting the observation noise covariance matrix parameters of the corresponding observations in the extended Kalman filter algorithm used in the fusion calculation based on the first weight, the second weight, and the third weight is as follows: In the standard extended Kalman filter algorithm, the observation noise covariance matrix Normally fixed, this embodiment uses the weight vector output in step S4 to dynamically reconstruct the matrix. Its mathematical structure is defined as follows: In this system, the diagonal elements (variance) of each submatrix are inversely proportional to their corresponding weights. The specific dynamic mapping function is as follows: In the formula: The nominal fundamental variance of each sensor (provided by the device datasheet) corresponds to the GNSS positioning module, IMU inertial navigation module, and visual recognition module, respectively. The first weight, second weight, and third weight are calculated in step S4; To prevent tiny positive numbers with a denominator of zero (e.g.) ); I is the identity matrix.

[0056] It should be noted that when When the (GNSS) weight decreases due to the signal-to-noise ratio, the denominator decreases, leading to... As the variance increases, the gain calculation formula for Kalman filtering is affected: middle, The increase leads to Kalman gain A decrease in noise level means the filter considers the observation to be extremely noisy and unreliable, thus reducing the correction magnitude for that observation and retaining more of the state equation's prediction (i.e., the inference value dominated by the IMU); conversely, when... When (visual weight) increases, the corresponding As the filter decreases, it will utilize more visual data to correct the state.

[0057] Preferably, the specific steps for using the adjusted observation noise covariance matrix parameters to perform state prediction and update, and output the three-dimensional coordinates and attitude information are as follows: This step executes the standard recursive loop of the extended Kalman filter: First, perform a time-updated prediction: using the optimal estimate from the previous time step. Using IMU data, the prior state estimate for the current moment is calculated. and prior error covariance ; Next, gain calculation is performed: based on the above-constructed... Matrix, calculate the optimal Kalman gain ; Then, a measurement update is performed: the observation residuals are calculated. And use gain to correct prior states: ; Finally, the posterior variance is updated: the error covariance matrix is ​​updated. This is used for iteration in the next time step.

[0058] Finally, from the updated posterior state vector The first three-dimensional components are extracted as three-dimensional coordinates, and the attitude components are extracted as precise attitude information, which are then output to the flight control system in real time at a frequency of over 100Hz.

[0059] In this embodiment, a deep fusion strategy that maps dynamic weights to a dynamic covariance matrix enables seamless soft switching of the UAV positioning system. Compared to the rigid logic of traditional algorithms that switch positioning systems based on GPS signal strength thresholds, this method achieves seamless soft switching by continuously adjusting... The size of the matrix allows the system to smoothly transition the positioning control from satellite navigation to vision and inertial navigation as the GNSS signal gradually deteriorates (e.g., as a drone slowly approaches a high-voltage line), preventing abrupt changes in positioning coordinates. The output three-dimensional coordinate accuracy can reach the centimeter level (better than ±3cm), and the attitude information has undergone cross-correction of multi-source data, effectively suppressing IMU drift and providing a stable spatiotemporal reference for the subsequent precise grasping and mounting of the robotic arm in a small space.

[0060] Step S6: Based on the three-dimensional coordinates and attitude information, control the drone to hover and adjust the attitude of the drone's robotic arm to install the bird deterrent device at the preset point on the target tower.

[0061] In a preferred embodiment, the robotic arm is a gravity-oriented flexible connection structure; the step of controlling the drone to hover and adjusting the robotic arm attitude of the drone based on the three-dimensional coordinates and attitude information specifically includes: controlling the flight position of the drone based on the three-dimensional coordinates, so that the drone hovers above a preset point on the target tower; and monitoring the instantaneous sway amplitude of the drone based on the three-dimensional coordinates, attitude information, and IMU inertial navigation data. When the swaying amplitude exceeds the preset safety range and causes the robotic arm to swing, the motor speed of the drone is controlled to perform reverse attitude compensation to actively suppress the swinging of the robotic arm.

[0062] Preferably, the specific steps for controlling the flight position of the UAV based on the three-dimensional coordinates, so that the UAV hovers above the preset point of the target tower, are as follows: First, the high-precision fusion positioning result output in step S5 is used as the feedback input for the flight control system. The flight control computer runs a cascaded PID control algorithm: the outer loop is the position loop and the inner loop is the attitude loop. Through this dual-loop control, even in mountainous areas with irregular gusts, the UAV can adjust the motor speed at a frequency of ≥50Hz to keep the aircraft in a safe hovering zone 2 to 3 meters directly above the target tower, strictly controlling the horizontal drift within ±3cm, and providing a relatively stationary inertial reference frame for subsequent robotic arm movements.

[0063] Preferably, the instantaneous sway amplitude of the UAV is monitored based on the three-dimensional coordinates, attitude information, and IMU inertial navigation data; when the sway amplitude exceeds a preset safety range causing the robotic arm to swing, the motor speed of the UAV is controlled to perform reverse attitude compensation, actively suppressing the swing of the robotic arm, specifically including: The flight control system reads the attitude information output in step S5 and the high-frequency angular velocity data from the IMU in real time. When it detects a sudden change in the horizontal acceleration of the UAV, and this change may excite the natural frequency oscillation of the robotic arm (e.g., the predicted swing amplitude exceeds ±3∘), the flight control system immediately intervenes and superimposes a reverse attitude compensation command while maintaining the position loop unchanged. For example, when it detects that the robotic arm has a tendency to swing to the left, the control system instantly adjusts the motor speed to generate a small, short-term leftward lateral acceleration in the UAV body, thereby moving the suspension point of the robotic arm to catch up with and counteract the kinetic energy of the pendulum below, and maintain the relative stability of the robotic arm.

[0064] In a preferred embodiment, when the drone hovers and tilts to resist wind loads, the gravity self-stabilizing characteristics of the flexible connection structure are utilized to passively deflect the robotic arm relative to the drone's fuselage, thereby maintaining the vertical attitude of the robotic arm's end effector. Specifically, this includes: This invention features a unique dual-axis attitude decoupling joint connected between the drone's underside and the top of the robotic arm. Utilizing the principle of gravity orientation, the robotic arm, which includes the electromagnetic adsorption mechanism and bird deterrent device, is treated as a physical pendulum. Since the total weight of the robotic arm and its load is much greater than the friction of the connecting joint, under the influence of gravitational torque, regardless of how the drone tilts to resist wind loads, the robotic arm can always passively find and align with the vertical line of gravity. This mechanism makes the robotic arm appear to "float" below the drone's fuselage, naturally isolating the drone's high-frequency attitude adjustment movements and ensuring that the installation end is always perpendicular to the ground plane.

[0065] In a preferred embodiment, the end of the robotic arm is provided with a variable-force electromagnetic adsorption mechanism. Installing the bird-repelling device at a preset point on the target pole specifically includes: adjusting and controlling the robotic arm's posture to push the bird-repelling device to the preset point and detecting the contact pressure in real time; when the contact pressure meets the installation conditions, controlling the electromagnetic adsorption mechanism to release the bird-repelling device so that the bird-repelling device is installed at the preset point on the target pole; wherein, the driving current of the electromagnetic adsorption mechanism is determined based on the load type information of the bird-repelling device.

[0066] Preferably, before installing the bird deterrent device, an automatic determination process for the driving current of the electromagnetic adsorption mechanism is included: the electromagnetic adsorption mechanism is equipped with a control module, which has a pre-stored lookup table to map the load type to the target adsorption force value (e.g., 50N for an ultrasonic bird deterrent and 80N for a bionic model); before the operation begins, the control module reads the tag information of the device to be mounted and retrieves the corresponding basic current value; during execution, the control module does not simply output a constant voltage, but adopts a current closed-loop PID control strategy, which uses a current sampling resistor to provide real-time feedback of the actual current and dynamically adjusts the duty cycle of the PWM wave to overcome the instability of the adsorption force caused by battery voltage fluctuations or coil heating, ensuring that the adsorption force is always stable within the ±2N accuracy range of the target value, thus guaranteeing the stability of the assembly process.

[0067] Preferably, the entire control of the robotic arm's posture, pushing the bird-repelling device to a preset position and detecting the contact pressure in real time, specifically includes: Once the drone carrying the bird deterrent device arrives above the target pole's crossarm, the flight control system switches to a force-position hybrid control mode. The drone controls its robotic arm to descend slowly while simultaneously monitoring the changes in the end-effector pressure sensor's values ​​in real time. When the pressure value abruptly changes from zero to a preset contact threshold (e.g., 5N), the system determines that the device's base has physically contacted the pole's crossarm. The drone then continues its slight descent, using its lift difference to generate a downward pushing force. During this process, the bird deterrent device's unique rectangular hook and elastic claw structure is compressed and contracted, causing the pressure sensor's reading to show a linear upward trend, ensuring a smooth transition from visual guidance to force confirmation.

[0068] Preferably, when the contact pressure meets the installation conditions, the electromagnetic adsorption mechanism is controlled to release the bird-repelling device, so that the bird-repelling device is installed at a preset point on the target tower, specifically including: When the electromagnetic adsorption mechanism detects that the pressure reading reaches the preset locking threshold (e.g., 20N-30N) and is accompanied by a momentary pressure unloading, it indicates that the elastic claw has successfully crossed the edge of the crossarm and rebounded to lock. Upon recognizing this signal, the electromagnetic adsorption mechanism immediately triggers a three-stage release logic: First, it cuts off the drive current of the electromagnetic adsorption mechanism; second, it briefly outputs a reverse pulse current to demagnetize and eliminate residual magnetic adhesion; third, it sends a feedback signal to the flight control system to control the drone to climb rapidly, so that the robotic arm is completely detached from the installed bird deterrent device in physical space, completing the installation task.

[0069] In this embodiment, the refined installation strategy integrating flexible passive orientation, active sway control, and force feedback is cleverly implemented. This gravity-oriented flexible connection structure, combined with motor reverse attitude compensation, cleverly solves the physical contradiction that the UAV cannot maintain the verticality of the robotic arm end when resisting crosswind tilt. This is equivalent to constructing a virtual image-stabilized gimbal structure to ensure the stability of the installation posture. At the same time, by constructing a load-current adaptive matching and PID closed-loop control mechanism, flexible compatibility with bird deterrent devices of different weights is achieved, avoiding operational risks caused by abnormal suction. Finally, the release logic based on the contact pressure change characteristics physically confirms the latch lock state, effectively preventing false installation risks caused by visual deception, and significantly improving the success rate and safety of unmanned operations in complex power environments.

[0070] like Figure 2 As shown, another embodiment of the present invention also provides a positioning and installation device for a bird deterrent device, including: a data acquisition module, a filtering processing module, a quality assessment module, a weight adjustment module, a fusion calculation module, and an installation control module; The data acquisition module is used to acquire GNSS positioning data, IMU inertial navigation data, and environmental image data collected by the visual sensor at the current flight location of the UAV. A filtering module is used to perform electromagnetic noise filtering on the GNSS positioning data to obtain first positioning data; The quality assessment module is used to calculate the signal-to-noise ratio of the first positioning data in real time, and to calculate the feature matching degree between the environmental image data and the preset target feature information; The weight adjustment module is used to determine, based on the signal-to-noise ratio and the feature matching degree, a first weight corresponding to the first positioning data, a second weight corresponding to the IMU inertial navigation data, and a third weight corresponding to the environmental image data; wherein, the first weight is positively correlated with the signal-to-noise ratio, the third weight is positively correlated with the feature matching degree, and the second weight is negatively correlated with the signal-to-noise ratio; The fusion computing module is used to perform fusion computing on the first positioning data, the IMU inertial navigation data and the environmental image data based on the first weight, the second weight and the third weight, using the extended Kalman filter algorithm, and output the three-dimensional coordinates and attitude information of the UAV. The installation control module is used to control the drone to hover and adjust the attitude of the drone's robotic arm based on the three-dimensional coordinates and attitude information, so as to install the bird deterrent device at the preset point of the target pole.

[0071] It is understood that the above-described device embodiments correspond to the method embodiments of the present invention, and can implement the positioning and installation method of the bird deterrent device provided by any of the above-described method embodiments of the present invention.

[0072] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can specifically be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0073] Based on the above-described embodiment of the positioning and installation method for a bird-repelling device, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements any of the positioning and installation methods for the bird-repelling device of the present invention.

[0074] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.

[0075] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0076] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0077] Based on the above-described method embodiments, another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the positioning and installation method of a bird-repelling device as described in any of the above-described method embodiments of the present invention.

[0078] The modules / units integrated in the device / terminal equipment, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0079] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for positioning and installing a bird-repelling device, characterized in that, include: Acquire GNSS positioning data, IMU inertial navigation data, and environmental image data collected by visual sensors at the current flight location of the UAV; The GNSS positioning data is subjected to electromagnetic noise filtering to obtain the first positioning data; The signal-to-noise ratio of the first positioning data is calculated in real time, and the feature matching degree between the environmental image data and the preset tower feature information is calculated. Based on the signal-to-noise ratio and the feature matching degree, a first weight corresponding to the first positioning data, a second weight corresponding to the IMU inertial navigation data, and a third weight corresponding to the environmental image data are determined respectively; wherein, the first weight is positively correlated with the signal-to-noise ratio, the third weight is positively correlated with the feature matching degree, and the second weight is negatively correlated with the signal-to-noise ratio; Based on the first weight, the second weight, and the third weight, the first positioning data, the IMU inertial navigation data, and the environmental image data are fused and calculated to output the three-dimensional coordinates and attitude information of the UAV. Based on the three-dimensional coordinates and attitude information, the drone is controlled to hover and the attitude of the drone's robotic arm is adjusted to install the bird deterrent device at the preset point on the target tower.

2. The positioning and installation method of the bird-repelling device as described in claim 1, characterized in that, The step of performing electromagnetic noise filtering on the GNSS positioning data to obtain first positioning data specifically includes: determining the target stopband frequency range based on the power line's characteristic frequency and harmonic frequency range; constructing a band-stop filter for the target stopband frequency range; using the band-stop filter to perform frequency domain filtering on the GNSS positioning data, suppressing signal components within the target stopband frequency range, and outputting the first positioning data after interference filtering.

3. The positioning and installation method of the bird-repelling device as described in claim 2, characterized in that, The calculation of the feature matching degree between the environmental image data and the preset tower feature information specifically includes: Construct a preset feature library containing a subset of feature point descriptors for key components of the target tower; extract local feature points within the current field of view based on the environmental image data; match the local feature points with the subset of feature point descriptors in the preset feature library and calculate the number of successfully matched feature points; map the number of successfully matched feature points to the feature matching degree.

4. The positioning and installation method of the bird-repelling device as described in claim 3, characterized in that, The step of determining a first weight corresponding to the first positioning data, a second weight corresponding to the IMU inertial navigation data, and a third weight corresponding to the environmental image data based on the signal-to-noise ratio and the feature matching degree specifically includes: Obtain the first baseline weight value of the first weight, the second baseline weight value of the second weight, and the third baseline weight value of the third weight; wherein the sum of the first baseline weight value, the second baseline weight value, and the third baseline weight value is 1; When the signal-to-noise ratio is greater than a preset first signal-to-noise ratio threshold, the first weight is increased from the first baseline weight value to the first dominant weight value, the second weight is decreased from the second baseline weight value to the first auxiliary weight value, and the third weight is decreased from the third baseline weight value to the second auxiliary weight value; wherein, the first auxiliary weight value and the second auxiliary weight value are both less than the first dominant weight value, and the sum of the first dominant weight value, the first auxiliary weight value, and the second auxiliary weight value is 1; When the signal-to-noise ratio (SNR) is between a preset first SNR threshold and a preset second SNR threshold, it is determined whether the feature matching degree meets the preset reliability condition. If yes, the third weight is increased from the third baseline weight value to the visual compensation weight value, the first weight is decreased from the first baseline weight value to the GNSS suppression weight value, and the second weight remains unchanged at the second baseline weight value. If no, the first weight, the second weight, and the third weight remain unchanged at their corresponding baseline weight values. The sum of the visual compensation weight value, the GNSS suppression weight value, and the second baseline weight value is 1. When the signal-to-noise ratio is less than the second signal-to-noise ratio threshold, the first weight is reduced from the first baseline weight value to zero, the second weight is increased from the second baseline weight value to the second dominant weight value, and the third weight is adjusted from the third baseline weight value to the visual correction weight value according to the feature matching degree; wherein, the visual correction weight value is positively correlated with the feature matching degree, and the sum of the visual correction weight value and the second dominant weight value is 1.

5. The positioning and installation method of the bird-repelling device as described in claim 4, characterized in that, The step of fusing and calculating the first positioning data, the IMU inertial navigation data, and the environmental image data according to the first weight, the second weight, and the third weight to output the three-dimensional coordinates and attitude information of the UAV specifically includes: constructing a state equation containing the UAV's position and velocity states, and an observation equation containing the first positioning data, the IMU inertial navigation data, and the environmental image data; dynamically adjusting the observation noise covariance matrix parameters of the corresponding observations in the extended Kalman filter algorithm used in the fusion calculation according to the first weight, the second weight, and the third weight; and using the adjusted observation noise covariance matrix parameters to perform state prediction and update, and output the three-dimensional coordinates and attitude information.

6. The positioning and installation method of the bird-repelling device as described in claim 5, characterized in that, Based on the three-dimensional coordinates and attitude information, the drone is controlled to hover and the attitude of its robotic arm is adjusted. Specifically, this includes: controlling the flight position of the drone based on the three-dimensional coordinates, so that the drone hovers above a preset point on the target tower; and monitoring the instantaneous sway amplitude of the drone based on the three-dimensional coordinates, attitude information, and IMU inertial navigation data. When the swaying amplitude exceeds the preset safety range and causes the robotic arm to swing, the motor speed of the drone is controlled to perform reverse attitude compensation to actively suppress the swinging of the robotic arm.

7. The positioning and installation method of the bird-repelling device as described in claim 6, characterized in that, The robotic arm is equipped with a variable-force electromagnetic adsorption mechanism at its end. Installing the bird-repelling device at a preset point on the target pole involves: adjusting and controlling the robotic arm's posture to push the bird-repelling device to the preset point and monitoring the contact pressure in real time; when the contact pressure meets the installation conditions, controlling the electromagnetic adsorption mechanism to release the bird-repelling device so that it is installed at the preset point on the target pole; wherein the driving current of the electromagnetic adsorption mechanism is determined based on the load type information of the bird-repelling device.

8. A positioning and installation device for a bird deterrent device, characterized in that, include: The system includes a data acquisition module, a filtering module, a quality assessment module, a weight adjustment module, a fusion calculation module, and an installation control module. The data acquisition module is used to acquire GNSS positioning data, IMU inertial navigation data, and environmental image data collected by the visual sensor at the current flight location of the UAV. A filtering module is used to perform electromagnetic noise filtering on the GNSS positioning data to obtain first positioning data; The quality assessment module is used to calculate the signal-to-noise ratio of the first positioning data in real time, and to calculate the feature matching degree between the environmental image data and the preset target feature information; The weight adjustment module is used to determine, based on the signal-to-noise ratio and the feature matching degree, a first weight corresponding to the first positioning data, a second weight corresponding to the IMU inertial navigation data, and a third weight corresponding to the environmental image data; wherein, the first weight is positively correlated with the signal-to-noise ratio, the third weight is positively correlated with the feature matching degree, and the second weight is negatively correlated with the signal-to-noise ratio; The fusion computing module is used to perform fusion computing on the first positioning data, the IMU inertial navigation data and the environmental image data based on the first weight, the second weight and the third weight, using the extended Kalman filter algorithm, and output the three-dimensional coordinates and attitude information of the UAV. The installation control module is used to control the drone to hover and adjust the attitude of the drone's robotic arm based on the three-dimensional coordinates and attitude information, so as to install the bird deterrent device at the preset point of the target pole.

9. A terminal device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements the positioning and installation method of the bird deterrent device as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, include: A stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the positioning and installation method of the bird deterrent device as described in any one of claims 1-7.

Images