Bridge unmanned aerial vehicle inspection positioning system and method based on embedded magnetic coding grid

Abstract

The invention discloses a bridge unmanned aerial vehicle routing inspection positioning system and method based on pre-embedded magnetic coding grids, and the system comprises a magnetic coding grid layer which is arranged in a bridge GNSS signal blind area, codes absolute coordinate beacons through the arrangement of permanent magnet poles of passive magnetic coding units, and carries a three-axis magnetometer on an airborne magnetic sensing layer to collect magnetic field data in real time. After interference compensation and signal identification processing, a magnetic field component is output, a positioning calculation layer receives the magnetic field component, a relative angle is calculated by adopting an attitude-independent magnetic induction vector algorithm, and absolute positioning coordinates of the unmanned aerial vehicle are output in combination with a known absolute coordinate beacon of a coding unit. And the disease mapping layer takes the absolute coordinates of the unmanned aerial vehicle and a disease image shot by the unmanned aerial vehicle as input, and associates and matches the two to complete automatic disease labeling of the bridge BIM model. According to the invention, centimeter-level absolute positioning of the unmanned aerial vehicle under the GNSS blind area at the bottom of the bridge is realized, and the problem of inspection and positioning of the bottom of the bridge is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of bridge structural defect inspection technology, and in particular to a bridge UAV inspection and positioning system and method based on a pre-embedded magnetic coding grid. Background Technology

[0002] As a critical node in transportation infrastructure, the structural safety of bridges directly impacts the smooth flow of traffic and the safety of people's lives and property. With the continuous increase in the total number of bridges and their service life, the need for bridge structural defect detection and maintenance is becoming increasingly urgent. Bridge surface defects mainly include cracks, exposed reinforcement, spalling, honeycomb-like pitting, and damage to expansion joints. Timely detection and accurate location of these defects are the foundation for bridge safety assessments and maintenance decisions.

[0003] In recent years, drone inspection technology has been widely used in the field of bridge defect detection due to its advantages of mobility, comprehensive field of view, and high operational efficiency. By equipping drones with sensors such as high-definition cameras and infrared thermal imagers, drones can quickly acquire images of the bridge structure and automatically detect defects using image recognition algorithms, greatly improving inspection efficiency.

[0004] However, drones face a key technical bottleneck in bridge inspection: the underside of bridges, around piers, and inside box girders are typical GNSS signal-denying environments. When drones fly into these areas, satellite positioning signals are completely lost, leading to the following technical problems: Positioning drift and cumulative error problem Currently, the mainstream solution is to use visual SLAM or laser SLAM for relative positioning. UAVs use onboard cameras or lidar to perceive environmental features, build environmental maps in real time, and estimate their own position. However, SLAM methods inherently suffer from cumulative error: as the flight distance increases, the positioning error gradually amplifies, failing to meet the requirement of centimeter-level precision in locating defects. When a UAV flies inside a box girder tens of meters long, the terminal positioning error can reach the meter level, making it impossible to accurately trace the location of defects.

[0005] Poor environmental adaptability The underside of bridges often suffers from insufficient lighting and severe shadows, and the surfaces of bridge piers may be dirty or covered with moss, all of which can lead to visual recognition failure. The repetitive internal structure and lack of texture in box girders make visual SLAM prone to losing lock. While laser SLAM is unaffected by lighting conditions, its performance degrades in environments with smoke, dust, or water mist, and its high equipment cost and power consumption limit the drone's endurance.

[0006] Dependence on active devices To address the GNSS blind zone positioning problem, some studies have proposed deploying pseudosatellites or UWB base stations. UAVs achieve positioning by receiving signals emitted by these active devices. However, this approach has significant drawbacks: pseudosatellites and UWB base stations require power and regular maintenance, making them impractical in environments like bridges where frequent maintenance is difficult; deployment costs are high, hindering widespread application in large-scale bridge inspections; and active devices also present issues such as signal interference and electromagnetic compatibility.

[0007] The problem of bridge structure interference with magnetic fields Bridge structures extensively utilize ferromagnetic materials such as steel reinforcement. According to the magnetic field mirror theory, these ferromagnetic materials generate an equivalent "mirror source" in space, distorting the original magnetic field of the embedded magnetic beacon. This interfering magnetic field is at the same frequency as the beacon's magnetic field, making it difficult to separate using conventional filtering methods and severely impacting magnetic positioning accuracy. Currently, there is a lack of effective compensation schemes for interference from bridge steel reinforcement.

[0008] The issue of the correlation between the location of the defect and the BIM model Existing drone inspection systems often only record the approximate location of the drone (such as GPS coordinates) or rely on manual annotations such as "10 meters south of pier No. 3" for descriptive location information. This positioning method cannot achieve accurate mapping of defects onto the bridge's BIM model, making it difficult to meet the development needs of digital bridge maintenance.

[0009] In summary, accurate UAV positioning in GNSS blind zones at the bottom of bridges is a pressing technical challenge in bridge defect inspection. Existing technologies have significant shortcomings in positioning accuracy, environmental adaptability, equipment maintenance, coordinate system alignment, and anti-interference capabilities. Therefore, there is an urgent need for a UAV inspection positioning system that can achieve absolute positioning in bridge blind zones, is unaffected by UAV attitude, has strong anti-interference capabilities, is passive and maintenance-free, and can seamlessly integrate with BIM models. Summary of the Invention

[0010] To address the shortcomings of current technology, this invention proposes a bridge UAV inspection and positioning system and method based on pre-embedded magnetic coded grids, thus solving the problems mentioned in the background technology.

[0011] To achieve the above objectives, the present invention provides the following technical solution: a bridge UAV inspection and positioning system based on a pre-embedded magnetic coded grid, comprising: The magnetic coding grid layer is used to pre-embed or lay in the GNSS signal blind zone on the bottom surface of bridges, the surface of bridge piers or the inner wall of box girders. It consists of multiple passive magnetic coding units with known and fixed positions. Each passive magnetic coding unit encodes the absolute three-dimensional coordinate information of its own deployment point in the GNSS signal blind zone through the magnetic pole arrangement of the internal permanent magnet material, forming an absolute coordinate beacon with a unique magnetic field fingerprint feature. The airborne magnetic sensing layer, which is mounted on the inspection drone, includes a three-axis magnetometer and a signal processing module. It is used to collect magnetic field data of the current position of the drone in real time. After the magnetic field data is processed by the interference compensation layer, the magnetic field component data corresponding to each passive magnetic coding unit is separated by the signal identification algorithm. The positioning calculation layer is used to receive the magnetic field component data output by the airborne magnetic sensing layer. It adopts a magnetic induction vector positioning algorithm that is independent of the sensor attitude to calculate the relative pitch angle and relative azimuth angle of the UAV relative to each passive magnetic coding unit. Combined with the known absolute coordinate beacons of the passive magnetic coding units, the absolute coordinates of the UAV are determined. The defect mapping layer associates bridge defect images captured by drones with the calculated absolute coordinates of the drones, enabling automatic labeling of defect locations on the bridge BIM model.

[0012] Furthermore, the positioning solution layer includes: The inner product calculation module is used to separate three magnetic field components with different frequencies from the received magnetic field component data. , , ; These are the separated constant magnetic field components or reference frequency components; The separated effective signal component or synchronization frequency component; These are the separated orthogonal magnetic field components; Will and Performing a vector dot product yields... ;Will and Performing a vector dot product yields... ;Will and Performing a vector dot product yields... ; for and scalar inner product value; for and scalar inner product value; for and scalar inner product value; The modulus calculation module is used to calculate the modulus value based on... , , The vector magnitudes of each magnetic field component are calculated using the vector magnitude formula. , , ;in, for Vector magnitude; for Vector magnitude; for Vector magnitude; Angle calculation module, used to calculate angles based on... , , as well as , , Solve for the relative pitch angle of the passive magnetic coding unit. The relative azimuth angle of the passive magnetic coding unit ; Given three passive magnetic coding units with known absolute coordinate beacons i=1,2,3, the drone's current position is ; For the first In the absolute three-dimensional coordinates of a passive magnetic encoding unit Axis coordinates; For the first In the absolute three-dimensional coordinates of a passive magnetic encoding unit Axis coordinates; For the first In the absolute three-dimensional coordinates of a passive magnetic encoding unit Axis coordinates; For drones Axis coordinates; For drones Axis coordinates; For drones Axis coordinates; According to the magnetic dipole model, each passive magnetic coding unit is located at the current position of the UAV. The magnetic field vector generated at that location is represented by the relative distance between the UAV and each passive magnetic coding unit. and a function of relative heading angle, which includes relative pitch angle. relative azimuth ; For drones relative to the first The relative pitch angle of each passive magnetic coding unit; For drones relative to the first The relative azimuth angles of the passive magnetic coding units; For drones and the first The relative distance between passive magnetic coding units; By simultaneously solving the inner product equations and the modulus equations of the three passive magnetic coding units, the relative pitch angle is constructed. Relative azimuth and relative distance The nonlinear equations were solved using a numerical iterative method to obtain the converged relative pitch angle. Relative azimuth and relative distance ; For the first The relative pitch angle of a passive magnetic coding unit after convergence; For the first The relative azimuth angle of the passive magnetic coding unit after convergence; For the first The relative distance between passive magnetic coding units after convergence; The inner product equation is the inner product of the magnetic field vectors of different passive magnetic coding units, expressed as a relative pitch angle. and relative azimuth The function is used to constrain unknown relative pitch angles. and relative azimuth ; The modulus equation expresses the magnitude of the magnetic field vector generated by each passive magnetic coding unit at the UAV as a function of relative distance. A function used to constrain unknown relative distances. ; based on , and as well as Calculate the absolute coordinates of the UAV ( , , ), In the absolute three-dimensional coordinates of the UAV Axis coordinates; In the absolute three-dimensional coordinates of the UAV Axis coordinates; In the absolute three-dimensional coordinates of the UAV Axis coordinates.

[0013] Furthermore, the processing procedure of the airborne magnetic sensing layer is as follows: Using a drone or handheld device equipped with a three-axis magnetometer for calibration, the GNSS signal blind zone of the bridge is scanned, and the magnetic field fingerprint characteristics of each passive magnetic coding unit and its corresponding absolute coordinate beacon are recorded to generate a magnetic field fingerprint map of the bottom of the bridge. Multiple calibration points were selected on the magnetic field fingerprint map at the bottom of the bridge, and the magnetic field data at each point were measured. The mirror source parameters were obtained by using an iterative compensation algorithm based on the absolute coordinate beacon. The airborne magnetic sensing layer also includes an interference compensation layer, used to compensate the measured magnetic field data based on the mirror source parameters to obtain compensated magnetic field data, specifically: After the inspection drone flies into the GNSS signal blind zone, the airborne triaxial magnetometer collects the measured magnetic field data in real time, and uses an iterative compensation algorithm based on the mirror source parameters to compensate the measured magnetic field data to obtain the compensated magnetic field data. The signal processing module uses a signal identification algorithm to separate the magnetic field component data corresponding to each passive magnetic coding unit from the compensated magnetic field data.

[0014] Furthermore, the specific process for obtaining the mirror source parameters is as follows: When constructing the magnetic field fingerprint map of the bridge bottom, n≥3 calibration points are selected, and the coordinates of the calibration points are ( , , ); For calibration points Axis coordinates; For calibration points Axis coordinates; For calibration points Axis coordinates; n is the number of calibration points selected; The magnetic field vector at each calibration point is measured using a calibration drone or handheld device equipped with a triaxial magnetometer. ; Let J be the measured magnetic field vector at the j-th calibration point; Based on the known absolute coordinate beacon The theoretical magnetic field at each calibration point under interference-free conditions was calculated using a magnetic dipole magnetic field calculation algorithm. The interfering magnetic field is: ; Let be the interference-free theoretical magnetic field vector at the j-th calibration point; For the j-th calibration point to The interference magnetic field vector at each mirror source; based on Iterative optimization calculations are performed using an iterative compensation algorithm. Iteratively derive the distance from the calibration point to the mirror source. Based on The estimated location of the mirror source is obtained by solving the three-dimensional spatial distance equation. ; For the j-th calibration point to the j-th calibration point Spatial distance between mirror sources; For the first A mirror source in a three-dimensional Cartesian coordinate system Axis coordinate estimates; For the first A mirror source in a three-dimensional Cartesian coordinate system Axis coordinate estimates; For the first A mirror source in a three-dimensional Cartesian coordinate system Axis coordinate estimates; Mirror sources have their own local coordinate system, utilizing Using the vector information, solve for the rotation matrix from the local coordinate system of the mirror source to the navigation coordinate system in the UAV's magnetic positioning and navigation mode; Location of mirror source The rotation matrix is ​​used as the mirror source parameter.

[0015] Furthermore, the specific arrangement of the passive magnetic coding units on the bridge bottom surface, pier surface, or inner wall of the box girder is as follows: Bridge underside: Passive magnetic coding units are pre-embedded or laid out on the bridge underside in an equidistant rectangular grid. The bridge pier surface includes cylindrical piers and square piers: the passive magnetic coding units are laid out in a circular + vertical grid pattern on the cylindrical piers; the passive magnetic coding units are laid out in a rectangular grid pattern on the square piers. Box girder inner wall: The passive magnetic coding units are pre-embedded in the inner wall of the box girder and arranged in a rectangular grid extending longitudinally or laterally along the inner wall of the box girder. They fit the curved surface of the inner wall of the box girder to achieve no dead angle coverage and completely cover the entire inner wall of the box girder.

[0016] Furthermore, the processing procedure for the disease mapping layer is as follows: The defect mapping layer associates the bridge defect images captured by the UAV with the calculated absolute coordinates of the UAV, and maps the defect pixels on the bridge defect images to the surface of the bridge BIM model using the ray casting method, thereby automatically marking the defect locations on the bridge BIM model.

[0017] Furthermore, the process for automatically annotating the locations of bridge defects in the bridge defect images onto the bridge BIM model is as follows: The drone flies along a preset route while simultaneously capturing images of bridge defects using a camera. For each image of the bridge defects captured, the flight control system records the drone's absolute coordinates. , , (and shooting posture, including relative pitch angle) and relative azimuth ; After the inspection is completed, the absolute coordinates of the drone will be recorded. , , The system imports the captured images of bridge defects into the bridge BIM management system, enabling defect localization from the captured images. This includes: Image recognition: Deep learning algorithms are used to identify cracks, exposed rebar, and spalling defects in bridge defect images to obtain the pixel coordinates of defects in the bridge defect images; defect pixels are then identified from the pixel coordinates of defects in the bridge defect images. Ray mapping: based on the absolute coordinates of the UAV ( , , Based on the shooting posture, construct a ray originating from the camera's optical center and passing through the defect pixels, and calculate the coordinates of the intersection point between the ray and the surface of the bridge BIM model; Coordinate transformation: Convert the coordinates of the intersection points into absolute coordinates in the bridge BIM model to enable automatic annotation of defects on the bridge BIM model; Report generation: After automatic annotation, an inspection report containing information on the type, size, and location of defects is generated.

[0018] A bridge UAV inspection and positioning method based on pre-embedded magnetic coded grids is applied to a bridge UAV inspection and positioning system based on pre-embedded magnetic coded grids, including: Step S1: Pre-embed or lay multiple passive magnetic coding units in the GNSS signal blind zone on the bottom surface of the bridge, the surface of the pier, or the inner wall of the box girder. Each passive magnetic coding unit encodes the absolute three-dimensional coordinate information of its own deployment point in the GNSS signal blind zone, forming an absolute coordinate beacon with a unique magnetic field fingerprint feature. Step S2: Using a calibration drone or handheld device equipped with a three-axis magnetometer, scan the GNSS signal blind zone of the bridge, record the magnetic field fingerprint characteristics of each passive magnetic coding unit and the absolute coordinate beacon corresponding to the passive magnetic coding unit, and generate a magnetic field fingerprint map of the bottom of the bridge. Step S21: Select multiple calibration points on the magnetic field fingerprint map at the bottom of the bridge, measure the magnetic field data at each calibration point, and calculate the magnetic field data using an iterative compensation algorithm based on the absolute coordinate beacon to obtain the mirror source parameters; Step S3: After the inspection drone flies into the GNSS signal blind zone, the onboard triaxial magnetometer collects the measured magnetic field data of the current position in real time, and compensates the measured magnetic field data based on the pre-obtained mirror source parameters to obtain the compensated magnetic field data; the signal processing module separates the magnetic field component data corresponding to each passive magnetic coding unit from the compensated magnetic field data through the signal identification algorithm. Step S4: Using a magnetic induction vector positioning algorithm that is independent of sensor attitude, the relative pitch and relative azimuth angles of the UAV relative to each passive magnetic coding unit are calculated based on the magnetic field component data. Combined with the absolute coordinate beacon, the absolute coordinates of the UAV are determined. Step S5: The drone flies along the preset route and takes pictures of bridge defects. The flight control system records the absolute coordinates and shooting attitude of the drone at the moment each bridge defect picture is taken. Step S6: Associate the bridge defect images with the absolute coordinates and shooting posture of the UAV, and use the ray casting method to map the defect pixels on the bridge defect images onto the surface of the bridge BIM model, so as to automatically mark the defect locations on the bridge BIM model.

[0019] An electronic device includes a processor, a memory, and a bus, wherein the processor and the memory are connected via the bus, wherein the memory is used to store a set of program codes, and the processor is used to call the program codes stored in the memory to execute a bridge UAV inspection and positioning method based on a pre-embedded magnetic coded grid.

[0020] A non-volatile computer storage medium storing computer-executable instructions, the computer being able to execute a bridge UAV inspection and positioning method based on a pre-embedded magnetic coded grid.

[0021] Compared with existing technologies, the present invention has the following advantages: This invention achieves centimeter-level absolute positioning of UAVs by deploying a magnetic coding grid layer in the GNSS signal blind zone of the bridge, eliminating cumulative positioning errors. It employs a magnetic induction vector positioning algorithm independent of sensor attitude, eliminating the influence of UAV attitude on positioning and ensuring strong positioning stability. Based on the magnetic field mirror theory, it effectively counteracts magnetic field interference from ferromagnetic materials such as bridge steel bars, improving positioning accuracy. The passive magnetic coding unit requires no power supply or regular maintenance, resulting in low deployment costs and strong adaptability. It can correlate bridge defect images with the absolute coordinates of the UAV, accurately mapping them to the bridge BIM model using the ray casting method, enabling automatic defect annotation and generating precise inspection reports, thus meeting the needs of digital bridge maintenance. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the overall architecture of the system of the present invention.

[0023] Figure 2 This is a schematic diagram of the layout of the magnetic coding unit on the bottom surface of the bridge according to the present invention.

[0024] Figure 3 This is a schematic diagram showing the layout of the magnetic coding units on the surface of the bridge pier according to the present invention.

[0025] Figure 4 This is a schematic diagram showing the layout of the magnetic coding unit on the inner wall of the box girder according to the present invention.

[0026] Figure 5 This is a schematic diagram illustrating the structure and magnetic pole arrangement of the passive magnetic encoding unit of the present invention.

[0027] Figure 6 This is a flowchart of the method of the present invention. Detailed Implementation

[0028] Example 1

[0029] like Figure 1 As shown, the present invention provides a technical solution: a bridge UAV inspection and positioning system based on a pre-embedded magnetic coded grid, comprising: The magnetic coding grid layer is used to pre-embed or lay in the GNSS signal blind zone on the bottom surface of bridges, the surface of bridge piers, or the inner wall of box girders. It consists of multiple passive magnetic coding units with known and fixed positions, which are arranged in a grid. Each passive magnetic coding unit encodes its absolute three-dimensional coordinate information of its placement point in the GNSS signal blind zone through the magnetic pole arrangement of its internal permanent magnet material, forming an absolute coordinate beacon with a unique magnetic field fingerprint. The airborne magnetic sensing layer, which is mounted on the inspection drone, includes a three-axis magnetometer and a signal processing module. It is used to collect magnetic field data of the current position of the drone in real time. After the magnetic field data is processed by the interference compensation layer, the magnetic field component data corresponding to each passive magnetic coding unit is separated by the signal identification algorithm. The positioning calculation layer is used to receive the magnetic field component data output by the airborne magnetic sensing layer. It adopts a magnetic induction vector positioning algorithm that is independent of the sensor attitude to calculate the relative pitch angle and relative azimuth angle of the UAV relative to each passive magnetic coding unit. Combined with the known absolute coordinate beacons of the passive magnetic coding units, the absolute coordinates of the UAV are determined. The defect mapping layer associates bridge defect images captured by drones with the calculated absolute coordinates of the drones, enabling automatic labeling of defect locations on the bridge BIM model.

[0030] The specific arrangement of the passive magnetic coding units on the bridge bottom surface, pier surface, or inner wall of the box girder is as follows: like Figure 2 As shown, the bottom surface of the bridge (bottom of the box girder / slab girder): Layout method: The layout method adopts pre-embedding (during bridge construction) or laying (later installation), with equidistant rectangular grids arranged, and the unit spacing is set to 0.5-2m (which can be adjusted as needed), and the grid is fine-tuned to adapt to the curved surface of the bridge bottom.

[0031] like Figure 3 As shown, the surface of the bridge pier (cylindrical / square column pier) Layout method: The layout method is to lay (attach to the outer wall of the pier), and the grid layout is adapted to different pier shapes: round piers: adopt a ring + vertical grid form, square piers: adopt a rectangular grid form.

[0032] like Figure 4 As shown, the inner wall of the box girder (inner walls on both sides / top of the box girder) Layout method: The pre-embedded (construction stage) layout method is adopted. For the inner walls of both sides and the inner wall of the top of the box girder, a rectangular grid extending along the longitudinal / transverse direction of the box girder is arranged to fit the curved surface of the inner wall of the box girder to achieve no dead angle coverage and can completely cover the entire inner wall of the box girder. At the same time, considering the narrow and long structural characteristics of the box girder, the grid spacing is reduced as needed to ensure positioning accuracy.

[0033] like Figure 5 As shown, each passive magnetic coding unit consists of a unit shell, an encapsulation layer, a permanent magnet array, and a mounting base. Figure 5 As shown in (a) in the text; Permanent magnet arrays with different magnetic pole arrangements can obtain fingerprint features corresponding to different magnetic fields, such as... Figure 5 As shown in (b); The permanent magnet array includes passive magnetic coding unit A, passive magnetic coding unit B, and passive magnetic coding unit C; The passive magnetic coding unit A adopts a column-oriented, alternating-row magnetic pole arrangement (top left N, top right S, bottom left N, bottom right S), which generates a symmetrical double-peak magnetic field fingerprint feature A. The passive magnetic coding unit B adopts a row-oriented, column-oriented alternating magnetic pole arrangement (top left N, top right N, bottom left S, bottom right S), which generates a magnetic field fingerprint feature B with symmetrical double valley values. The passive magnetic coding unit C adopts a diagonally aligned magnetic pole arrangement (top left S, top right N, bottom left N, bottom right S), which generates an asymmetric single-peak and single-valley magnetic field fingerprint feature C.

[0034] The processing procedure of the airborne magnetic sensing layer is as follows: Using a drone or handheld device equipped with a three-axis magnetometer for calibration, the GNSS signal blind zone of the bridge is scanned, and the magnetic field fingerprint characteristics of each passive magnetic coding unit and the absolute coordinate beacon corresponding to the passive magnetic coding unit are recorded to generate a magnetic field fingerprint map of the bottom of the bridge. Multiple calibration points are selected on the magnetic field fingerprint map of the bottom of the bridge, and the magnetic field data at each point is measured. Based on the absolute coordinate beacon, an iterative compensation algorithm is used to calculate and obtain the mirror source parameters. The airborne magnetic sensing layer also includes an interference compensation layer, used to compensate the measured magnetic field data based on the mirror source parameters to obtain compensated magnetic field data, specifically: After the inspection drone flies into the GNSS signal blind zone, the airborne triaxial magnetometer collects the measured magnetic field data of the current location in real time, and uses an iterative compensation algorithm based on the mirror source parameters to compensate the measured magnetic field data to obtain the compensated magnetic field data. The signal processing module uses a signal identification algorithm to separate the magnetic field component data corresponding to each passive magnetic coding unit from the compensated magnetic field data.

[0035] The positioning solution layer includes: The inner product calculation module is used to separate three magnetic field components with different frequencies from the received magnetic field component data. , , ; These are the separated constant magnetic field components or reference frequency components; The separated effective signal component or synchronization frequency component; These are the separated orthogonal magnetic field components; Calculate the inner product value between each magnetic field component. , , ; for and scalar inner product value; for and scalar inner product value; for and scalar inner product value; The modulus calculation module is used to calculate the modulus value based on... , , The vector magnitudes of each magnetic field component are calculated using the formula for vector magnitude (Euclidean norm). , , ;in, for Vector magnitude; for Vector magnitude; for Vector magnitude; Angle calculation module, used to calculate angles based on... , , as well as , , Solve for the relative pitch angle of the passive magnetic coding unit. The relative azimuth angle of the passive magnetic coding unit , is represented as: ; ; ; In the formula, It is the sine value; The value is the cosine. The sensor attitude-independent magnetic induction vector positioning algorithm is based on the magnetic dipole model. , , The calculation is performed. Since the attitude of the UAV is unknown, it is difficult to directly solve the coordinates using the direction of the magnetic field vector. This invention eliminates the attitude influence by calculating the inner product and magnitude between vectors. The inner product and magnitude are rotationally invariant and are independent of the measurement coordinate system.

[0036] Given three passive magnetic coding units with known absolute coordinate beacons i=1,2,3, the drone's current position is ; For the first In the absolute three-dimensional coordinates of a passive magnetic encoding unit Axis coordinates; For the first In the absolute three-dimensional coordinates of a passive magnetic encoding unit Axis coordinates; For the first In the absolute three-dimensional coordinates of a passive magnetic encoding unit Axis coordinates; For drones Axis coordinates; For drones Axis coordinates; For drones Axis coordinates; According to the magnetic dipole model, each passive magnetic coding unit is located at the current position of the UAV. The magnetic field vector generated at that location is represented by the relative distance between the UAV and each passive magnetic coding unit. and a function of relative heading angle, which includes relative pitch angle. relative azimuth ; For drones relative to the first The relative pitch angle of each passive magnetic coding unit; For drones relative to the first The relative azimuth angles of the passive magnetic coding units; For drones and the first The relative distance between passive magnetic coding units; By simultaneously solving the inner product equations and the modulus equations of the three passive magnetic coding units, the relative pitch angle is constructed. relative azimuth and relative distance The nonlinear equations were solved using a numerical iterative method to obtain the converged relative pitch angle. Relative azimuth and relative distance ; For the first The relative pitch angle of a passive magnetic coding unit after convergence; For the first The relative azimuth angle of the passive magnetic coding unit after convergence; For the first The relative distance between passive magnetic coding units after convergence.

[0037] The inner product equation is the inner product of the magnetic field vectors of different passive magnetic coding units, expressed as a relative pitch angle. and relative azimuth The function is used to constrain unknown relative pitch angles. and relative azimuth .

[0038] The modulus equation expresses the magnitude of the magnetic field vector generated by each passive magnetic coding unit at the UAV as a function of relative distance. A function used to constrain unknown relative distances. .

[0039] based on , and as well as Calculate the absolute coordinates of the UAV ( , , ), represented as: ; ; ; In the formula, In the absolute three-dimensional coordinates of the UAV Axis coordinates; In the absolute three-dimensional coordinates of the UAV Axis coordinates; In the absolute three-dimensional coordinates of the UAV Axis coordinates.

[0040] absolute coordinates of the drone ( , , )and , The geometric correspondence is as follows: , ; In the formula, It is the arctangent function.

[0041] The processing procedure for the disease mapping layer is as follows: The defect mapping layer associates the bridge defect images captured by the UAV with the calculated absolute coordinates of the UAV, and maps the defect pixels on the bridge defect images to the surface of the bridge BIM model using the ray casting method, thereby automatically marking the defect locations on the bridge BIM model.

[0042] like Figure 6 As shown, the bridge UAV inspection and positioning method based on pre-embedded magnetic coded grids, applied to the bridge UAV inspection and positioning system based on pre-embedded magnetic coded grids, includes: Step S1: Pre-embed or lay multiple passive magnetic coding units in the GNSS signal blind zone on the bottom surface of the bridge, the surface of the pier, or the inner wall of the box girder. Each passive magnetic coding unit encodes the absolute three-dimensional coordinate information of its own deployment point in the GNSS signal blind zone, forming an absolute coordinate beacon with a unique magnetic field fingerprint feature. Step S2: Using a calibration drone or handheld device equipped with a three-axis magnetometer, scan the GNSS signal blind zone of the bridge, record the magnetic field fingerprint characteristics of each passive magnetic coding unit and the absolute coordinate beacon corresponding to the passive magnetic coding unit, and generate a magnetic field fingerprint map of the bottom of the bridge. Step S21: Select multiple calibration points on the magnetic field fingerprint map at the bottom of the bridge, measure the magnetic field data at each calibration point, and calculate the magnetic field data using an iterative compensation algorithm based on the absolute coordinate beacon to obtain the mirror source parameters; Step S3: After the inspection drone flies into the GNSS signal blind zone, the onboard triaxial magnetometer collects the measured magnetic field data of the current position in real time, and compensates the measured magnetic field data based on the pre-obtained mirror source parameters to obtain the compensated magnetic field data; the signal processing module separates the magnetic field component data corresponding to each passive magnetic coding unit from the compensated magnetic field data through the signal identification algorithm. Step S4: Using a magnetic induction vector positioning algorithm that is independent of sensor attitude, the relative pitch and relative azimuth angles of the UAV relative to each passive magnetic coding unit are calculated based on the magnetic field component data. Combined with the absolute coordinate beacon, the absolute coordinates of the UAV are determined. Step S5: The drone flies along the preset route and takes pictures of bridge defects. The flight control system records the absolute coordinates and shooting attitude of the drone at the moment each bridge defect picture is taken. Step S6: Associate the bridge defect images with the absolute coordinates and shooting posture of the UAV, and use the ray casting method to map the defect pixels on the bridge defect images onto the surface of the bridge BIM model, so as to automatically mark the defect locations on the bridge BIM model.

[0043] Example 2

[0044] The specific process for obtaining the mirror source parameters is as follows: When constructing the magnetic field fingerprint map of the bridge bottom, n≥3 calibration points are selected, and the coordinates of the calibration points are ( , , ); For calibration points Axis coordinates; For calibration points Axis coordinates; For calibration points Axis coordinates; n is the number of calibration points selected; The magnetic field vector at each calibration point is measured using a calibration drone or handheld device equipped with a triaxial magnetometer. ; Let J be the measured magnetic field vector at the j-th calibration point; Based on the known absolute coordinate beacon The theoretical magnetic field at each calibration point under interference-free conditions was calculated using a magnetic dipole magnetic field calculation algorithm. The interfering magnetic field is: ; Let be the interference-free theoretical magnetic field vector at the j-th calibration point; For the j-th calibration point to The interference magnetic field vector at each mirror source; According to the magnetic field mirror theory, the interfering magnetic field is equivalent to a virtual "mirror source". This can be addressed by solving an optimization problem: based on Iterative optimization calculations are performed using an iterative compensation algorithm. Iteratively derive the distance from the calibration point to the mirror source. Based on The estimated location of the mirror source is obtained by solving the three-dimensional spatial distance equation. , is represented as: ; In the formula, To obtain the minimum value; For the j-th calibration point to the j-th calibration point Spatial distance between mirror sources; For the first A mirror source in a three-dimensional Cartesian coordinate system Axis coordinate estimates; For the first A mirror source in a three-dimensional Cartesian coordinate system Axis coordinate estimates; For the first A mirror source in a three-dimensional Cartesian coordinate system Axis coordinate estimates; For the j-th calibration point Axis coordinates; For the j-th calibration point Axis coordinates; For the j-th calibration point Axis coordinates; a mirror source has its own local coordinate system, utilizing... Given the vector information, solve for the rotation matrix from the local coordinate system of the mirror source to the navigation coordinate system in the UAV's magnetic positioning and navigation mode.

[0045] Location of mirror source The rotation matrix is ​​used as the mirror source parameter; The navigation coordinate system is a global reference coordinate system (such as the NE-G coordinate system or NE-C coordinate system) used by the UAV inspection system to uniformly represent the positions of the UAV, passive magnetic coding unit, and mirror source. With attitude information.

[0046] Example 3

[0047] The process for automatically annotating the locations of bridge defects in images onto the bridge BIM model is as follows: The drone flies along a preset route while simultaneously capturing images of bridge defects using a camera. For each image of the bridge defects captured, the flight control system records the drone's absolute coordinates. , , (and shooting posture, including relative pitch angle) and relative azimuth ; After the inspection is completed, the absolute coordinates of the drone will be recorded. , , The system imports the captured images of bridge defects into the bridge BIM management system, enabling defect localization from the captured images. This includes: Image recognition: Deep learning algorithms are used to identify cracks, exposed rebar, and spalling defects in bridge defect images to obtain the pixel coordinates of defects in the bridge defect images; defect pixels are then identified from the pixel coordinates of defects in the bridge defect images. Ray mapping: based on the absolute coordinates of the UAV ( , , Based on the shooting posture, construct a ray from the camera's optical center through the defect pixels, and calculate the coordinates of the intersection point between the ray and the surface of the bridge BIM model (3D model). Coordinate transformation: Convert the coordinates of the intersection points into absolute coordinates in the bridge BIM model to enable automatic annotation of defects on the bridge BIM model; Report generation: After automatic annotation, an inspection report containing information on the type, size, and precise location of the defects is generated.

[0048] Example 4

[0049] An electronic device includes a processor, a memory, and a bus, wherein the processor and the memory are connected via the bus, wherein the memory is used to store a set of program codes, and the processor is used to call the program codes stored in the memory to execute a bridge UAV inspection and positioning method based on a pre-embedded magnetic coded grid.

[0050] Example 5

[0051] A non-volatile computer storage medium stores computer-executable instructions that execute a bridge UAV inspection and positioning method based on a pre-embedded magnetic coded grid.

[0052] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A bridge UAV inspection and positioning system based on a pre-embedded magnetic coded grid, characterized in that, include: The magnetic coding grid layer is used to pre-embed or lay in the GNSS signal blind zone on the bottom surface of bridges, the surface of bridge piers or the inner wall of box girders. It consists of multiple passive magnetic coding units with known and fixed positions. Each passive magnetic coding unit encodes the absolute three-dimensional coordinate information of its own deployment point in the GNSS signal blind zone through the magnetic pole arrangement of the internal permanent magnet material, forming an absolute coordinate beacon with a unique magnetic field fingerprint feature. The airborne magnetic sensing layer, which is mounted on the inspection drone, includes a three-axis magnetometer and a signal processing module. It is used to collect magnetic field data of the current position of the drone in real time. After the magnetic field data is processed by the interference compensation layer, the magnetic field component data corresponding to each passive magnetic coding unit is separated by the signal identification algorithm. The positioning calculation layer is used to receive the magnetic field component data output by the airborne magnetic sensing layer. It adopts a magnetic induction vector positioning algorithm that is independent of the sensor attitude to calculate the relative pitch angle and relative azimuth angle of the UAV relative to each passive magnetic coding unit. Combined with the known absolute coordinate beacons of the passive magnetic coding units, the absolute coordinates of the UAV are determined. The defect mapping layer associates bridge defect images captured by drones with the calculated absolute coordinates of the drones, enabling automatic labeling of defect locations on the bridge BIM model.

2. The bridge UAV inspection and positioning system based on pre-embedded magnetic coded grid as described in claim 1, characterized in that: The positioning solution layer includes: The inner product calculation module is used to separate three magnetic field components with different frequencies from the received magnetic field component data. , , ; These are the separated constant magnetic field components or reference frequency components; The separated effective signal component or synchronization frequency component; These are the separated orthogonal magnetic field components; Will and Performing a vector dot product yields... ;Will and Performing a vector dot product yields... ;Will and Performing a vector dot product yields... ; for and scalar inner product value; for and scalar inner product value; for and scalar inner product value; The modulus calculation module is used to calculate the modulus value based on... , , The vector magnitudes of each magnetic field component are calculated using the vector magnitude formula. , , ;in, for Vector magnitude; for Vector magnitude; for Vector magnitude; Angle calculation module, used to calculate angles based on... , , as well as , , Solve for the relative pitch angle of the passive magnetic coding unit. The relative azimuth angle of the passive magnetic coding unit ; Given three passive magnetic coding units with known absolute coordinate beacons i=1,2,3, the drone's current position is ; For the first In the absolute three-dimensional coordinates of a passive magnetic encoding unit Axis coordinates; For the first In the absolute three-dimensional coordinates of a passive magnetic encoding unit Axis coordinates; For the first In the absolute three-dimensional coordinates of a passive magnetic encoding unit Axis coordinates; For drones Axis coordinates; For drones Axis coordinates; For drones Axis coordinates; According to the magnetic dipole model, each passive magnetic coding unit is located at the current position of the UAV. The magnetic field vector generated at that location is represented by the relative distance between the UAV and each passive magnetic coding unit. and a function of relative heading angle, which includes relative pitch angle. relative azimuth ; For drones relative to the first The relative pitch angle of each passive magnetic coding unit; For drones relative to the first The relative azimuth angles of the passive magnetic coding units; For drones and the first The relative distance between passive magnetic coding units; By simultaneously solving the inner product equations and the modulus equations of the three passive magnetic coding units, the relative pitch angle is constructed. Relative azimuth and relative distance The nonlinear equations were solved using a numerical iterative method to obtain the converged relative pitch angle. Relative azimuth and relative distance ; For the first The relative pitch angle of a passive magnetic coding unit after convergence; For the first The relative azimuth angle of the passive magnetic coding unit after convergence; For the first The relative distance between passive magnetic coding units after convergence; The inner product equation is the inner product of the magnetic field vectors of different passive magnetic coding units, expressed as a relative pitch angle. and relative azimuth The function is used to constrain unknown relative pitch angles. and relative azimuth ; The modulus equation expresses the magnitude of the magnetic field vector generated by each passive magnetic coding unit at the UAV as a function of relative distance. A function used to constrain unknown relative distances. ; based on , and as well as Calculate the absolute coordinates of the UAV ( , , ), In the absolute three-dimensional coordinates of the UAV Axis coordinates; In the absolute three-dimensional coordinates of the UAV Axis coordinates; In the absolute three-dimensional coordinates of the UAV Axis coordinates.

3. The bridge UAV inspection and positioning system based on pre-embedded magnetic coded grid according to claim 2, characterized in that: The processing procedure of the airborne magnetic sensing layer is as follows: Using a drone or handheld device equipped with a three-axis magnetometer for calibration, the GNSS signal blind zone of the bridge is scanned, and the magnetic field fingerprint characteristics of each passive magnetic coding unit and its corresponding absolute coordinate beacon are recorded to generate a magnetic field fingerprint map of the bottom of the bridge. Multiple calibration points were selected on the magnetic field fingerprint map at the bottom of the bridge, and the magnetic field data at each point were measured. The mirror source parameters were obtained by using an iterative compensation algorithm based on the absolute coordinate beacon. The airborne magnetic sensing layer also includes an interference compensation layer, used to compensate the measured magnetic field data based on the mirror source parameters to obtain compensated magnetic field data, specifically: After the inspection drone flies into the GNSS signal blind zone, the airborne triaxial magnetometer collects the measured magnetic field data in real time, and uses an iterative compensation algorithm based on the mirror source parameters to compensate the measured magnetic field data to obtain the compensated magnetic field data. The signal processing module uses a signal identification algorithm to separate the magnetic field component data corresponding to each passive magnetic coding unit from the compensated magnetic field data.

4. The bridge UAV inspection and positioning system based on pre-embedded magnetic coded grid according to claim 3, characterized in that: The specific process for obtaining the mirror source parameters is as follows: When constructing the magnetic field fingerprint map of the bridge bottom, n≥3 calibration points are selected, and the coordinates of the calibration points are ( , , ); For calibration points Axis coordinates; For calibration points Axis coordinates; For calibration points Axis coordinates; n is the number of calibration points selected; The magnetic field vector at each calibration point is measured using a calibration drone or handheld device equipped with a triaxial magnetometer. ; Let J be the measured magnetic field vector at the j-th calibration point; Based on the known absolute coordinate beacon The theoretical magnetic field at each calibration point under interference-free conditions was calculated using a magnetic dipole magnetic field calculation algorithm. The interfering magnetic field is: ; Let be the interference-free theoretical magnetic field vector at the j-th calibration point; For the j-th calibration point to The interference magnetic field vector at each mirror source; based on Iterative optimization calculations are performed using an iterative compensation algorithm. Iteratively derive the distance from the calibration point to the mirror source. Based on The estimated location of the mirror source is obtained by solving the three-dimensional spatial distance equation. ; For the j-th calibration point to the j-th calibration point Spatial distance between mirror sources; For the first A mirror source in a three-dimensional Cartesian coordinate system Axis coordinate estimates; For the first A mirror source in a three-dimensional Cartesian coordinate system Axis coordinate estimates; For the first A mirror source in a three-dimensional Cartesian coordinate system Axis coordinate estimates; Mirror sources have their own local coordinate system, utilizing Using the vector information, solve for the rotation matrix from the local coordinate system of the mirror source to the navigation coordinate system in the UAV's magnetic positioning and navigation mode; Location of mirror source The rotation matrix is ​​used as the mirror source parameter.

5. The bridge UAV inspection and positioning system based on pre-embedded magnetic coded grid according to claim 4, characterized in that: The specific arrangement of passive magnetic coding units on the bridge bottom surface, pier surface, or inner wall of box girder is as follows: Bridge underside: Passive magnetic coding units are pre-embedded or laid out on the bridge underside in an equidistant rectangular grid. The bridge pier surface includes cylindrical piers and square piers: the passive magnetic coding units are laid out in a circular + vertical grid pattern on the cylindrical piers; the passive magnetic coding units are laid out in a rectangular grid pattern on the square piers. Box girder inner wall: The passive magnetic coding units are pre-embedded in the inner wall of the box girder and arranged in a rectangular grid extending longitudinally or laterally along the inner wall of the box girder. They fit the curved surface of the inner wall of the box girder to achieve no dead angle coverage and completely cover the entire inner wall of the box girder.

6. The bridge UAV inspection and positioning system based on pre-embedded magnetic coded grid according to claim 5, characterized in that: The processing procedure for the disease mapping layer is as follows: The defect mapping layer associates the bridge defect images captured by the UAV with the calculated absolute coordinates of the UAV, and maps the defect pixels on the bridge defect images to the surface of the bridge BIM model using the ray casting method, thereby automatically marking the defect locations on the bridge BIM model.

7. The bridge UAV inspection and positioning system based on pre-embedded magnetic coded grid according to claim 6, characterized in that: The process for automatically annotating the locations of bridge defects in images onto the bridge BIM model is as follows: The drone flies along a preset route while simultaneously capturing images of bridge defects using a camera. For each image of the bridge defects captured, the flight control system records the drone's absolute coordinates. , , (and shooting posture, including relative pitch angle) and relative azimuth ; After the inspection is completed, the absolute coordinates of the drone will be recorded. , , The system imports the captured images of bridge defects into the bridge BIM management system, enabling defect localization from the captured images. This includes: Image recognition: Deep learning algorithms are used to identify cracks, exposed rebar, and spalling defects in bridge defect images to obtain the pixel coordinates of defects in the bridge defect images; defect pixels are then identified from the pixel coordinates of defects in the bridge defect images. Ray mapping: based on the absolute coordinates of the UAV ( , , Based on the shooting posture, construct a ray originating from the camera's optical center and passing through the defect pixels, and calculate the coordinates of the intersection point between the ray and the surface of the bridge BIM model; Coordinate transformation: Convert the coordinates of the intersection points into absolute coordinates in the bridge BIM model to enable automatic annotation of defects on the bridge BIM model; Report generation: After automatic annotation, an inspection report containing information on the type, size, and location of defects is generated.

8. A bridge UAV inspection and positioning method based on a pre-embedded magnetic coded grid, applied to the bridge UAV inspection and positioning system based on a pre-embedded magnetic coded grid as described in any one of claims 1-7, characterized in that, include: Step S1: Pre-embed or lay multiple passive magnetic coding units in the GNSS signal blind zone on the bottom surface of the bridge, the surface of the pier, or the inner wall of the box girder. Each passive magnetic coding unit encodes the absolute three-dimensional coordinate information of its own deployment point in the GNSS signal blind zone, forming an absolute coordinate beacon with a unique magnetic field fingerprint feature. Step S2: Using a calibration drone or handheld device equipped with a three-axis magnetometer, scan the GNSS signal blind zone of the bridge, record the magnetic field fingerprint characteristics of each passive magnetic coding unit and the absolute coordinate beacon corresponding to the passive magnetic coding unit, and generate a magnetic field fingerprint map of the bottom of the bridge. Step S21: Select multiple calibration points on the magnetic field fingerprint map at the bottom of the bridge, measure the magnetic field data at each calibration point, and calculate the magnetic field data using an iterative compensation algorithm based on the absolute coordinate beacon to obtain the mirror source parameters; Step S3: After the inspection drone flies into the GNSS signal blind zone, the onboard triaxial magnetometer collects the measured magnetic field data of the current position in real time, and compensates the measured magnetic field data based on the pre-obtained mirror source parameters to obtain the compensated magnetic field data; the signal processing module separates the magnetic field component data corresponding to each passive magnetic coding unit from the compensated magnetic field data through the signal identification algorithm. Step S4: Using a magnetic induction vector positioning algorithm that is independent of sensor attitude, the relative pitch and relative azimuth angles of the UAV relative to each passive magnetic coding unit are calculated based on the magnetic field component data. Combined with the absolute coordinate beacon, the absolute coordinates of the UAV are determined. Step S5: The drone flies along the preset route and takes pictures of bridge defects. The flight control system records the absolute coordinates and shooting attitude of the drone at the moment each bridge defect picture is taken. Step S6: Associate the bridge defect images with the absolute coordinates and shooting posture of the UAV, and use the ray casting method to map the defect pixels on the bridge defect images onto the surface of the bridge BIM model, so as to automatically mark the defect locations on the bridge BIM model.

9. An electronic device, characterized in that, The system includes a processor, a memory, and a bus. The processor and the memory are connected via the bus. The memory stores a set of program code, and the processor calls the program code stored in the memory to execute the bridge UAV inspection and positioning method based on pre-embedded magnetic coded grid as described in claim 8.

10. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer can execute instructions to perform the bridge UAV inspection and positioning method based on pre-embedded magnetic coded grid as described in claim 8.

Images