Unmanned aerial vehicle wind measurement system and method

Abstract

The system comprises an ultrasonic anemometer, a redundant GPS (Global Positioning System), a height accelerometer, a high-precision accelerometer with three course angles, a CCD (Charge Coupled Device) industrial camera, a camera optical center, a high-precision geographic identifier, a ground host, an airborne host and an unmanned aerial vehicle platform, a high-precision geographic identification database is input into the system; the unmanned aerial vehicle platform carries an airborne host, an ultrasonic anemometer, a redundant GPS, a height accelerometer, a high-precision accelerometer and a CCD industrial camera. The unmanned aerial vehicle climbs and flies to the position above the identification, a CCD industrial camera is used for photographing the position above the identification, two or more geographic identifications with the enough distance are shot, and the unmanned aerial vehicle flies horizontally as much as possible; iteratively solving the current optical center position of the unmanned aerial vehicle; solving the navigational speed of the unmanned aerial vehicle; solving the navigational speed of the ultrasonic anemometer; the wind speed is solved, and wind following sailing is used for reducing measurement errors. According to the invention, high-precision and high-temporal-spatial-resolution wind field measurement can be realized.

Description

Technical Field

[0001] This invention relates to the field of wind measurement technology for unmanned aerial vehicles (UAVs), and more particularly to a wind measurement system and method for UAVs. Background Technology

[0002] Drones can be equipped with 2 / 3D ultrasonic anemometers or other wind measuring instruments, and like ground stations, they are contact-based measurements. The data obtained is accurate, and the sampling frequency can reach hundreds of Hz, which is far higher than the sampling speed of lidar (1~10 Hz). At the same time, drones can simultaneously measure other meteorological elements at high altitudes. However, drone wind measurement faces huge technical challenges, mainly including: (1) the positioning and motion state of drones: including three-dimensional position and attitude. Drones are different from ground stations, towers or radars, and they are in motion; (2) the flow generated by drones themselves can cause interference to the background wind field.

[0003] Classic UAV wind measurement techniques include analytical wind measurement, horizontal airspeed zeroing method, and maneuvering flight method. These wind measurement methods require the aircraft to fly at least 200-300 meters and take about 10 seconds. The relatively coarse spatiotemporal resolution can meet general meteorological requirements, but cannot meet industrial technical requirements. Summary of the Invention

[0004] Purpose of the invention: This invention provides a wind measurement system and method for unmanned aerial vehicles (UAVs), which can achieve high-precision and high spatiotemporal resolution wind field measurement.

[0005] Technical Solution: The present invention discloses a UAV wind measurement system, comprising: an ultrasonic anemometer or other wind measurement device, redundant GPS and altimeter, a high-precision accelerometer with three heading angles, a CCD industrial camera, a camera optical center, high-precision geotags, a ground host, an airborne host, and a UAV platform; the system inputs a high-precision geotag database, and the UAV platform carries an airborne host, an ultrasonic anemometer, redundant GPS and altimeter, a high-precision accelerometer, and a CCD industrial camera; the UAV climbs to a height above the geotags, takes a picture with the CCD industrial camera above the geotags, capturing geotags with at least two sufficiently spaced geotags, and the UAV flies as horizontally as possible; the current optical center position of the UAV is iteratively calculated, the UAV's speed is calculated, the speed of the ultrasonic anemometer is calculated, and the wind speed is calculated, using wind-following navigation to reduce measurement errors.

[0006] Accordingly, a method for wind measurement using a drone includes the following steps:

[0007] Step 1: Input the high-precision geographic identification database into the system;

[0008] Step 2: The UAV platform is equipped with an onboard host, an ultrasonic anemometer, redundant GPS and an altitude accelerometer, a high-precision accelerometer with three heading angles, and a CCD industrial camera;

[0009] Step 3: The drone climbs and flies to a height near the marker. It takes a picture above the marker with a CCD industrial camera, capturing at least two geographical markers with sufficient spacing. The drone should fly as horizontally as possible.

[0010] Step 4: Iteratively solve for the current optical center position of the UAV;

[0011] Step 5: Determine the speed of the UAV;

[0012] Step 6: Determine the speed of the aircraft using the ultrasonic anemometer;

[0013] Step 7: Calculate the wind speed to reduce measurement error.

[0014] Furthermore, in step 1, the system inputs a high-precision geoidentification database, including GPS information and image information, which are two-dimensional or three-dimensional. The high-precision geoidentification database is simultaneously and collaboratively stored in the ground host and the airborne host.

[0015] Furthermore, in step 3, the drone climbs and flies above the marker, where a CCD industrial camera takes a picture. The drone flies as horizontally as possible to ensure the image is in focus.

[0016] (1)

[0017] (2)

[0018] Where f is the focal length, u is the object distance, and v is the phase distance. For a fixed-focus f-lens and a fixed distance v, the object distance u for precise focusing and the magnification are... It is definite; in the camera coordinate system, the optical center is the origin, the X-axis of the CCD industrial camera is the X-axis, and the Z-axis is outward along the optical axis. The lens mapping law is written as... =- = , among which the target location Let be the position vector of the target geographic marker relative to the optical center; the CCD industrial camera continuously captures continuous images of a relatively long line connecting the current geographic marker at high speed, and obtains the optimal focused image according to the Fourier optical imaging transform and the magnification of the line connecting the two markers; within the depth of field, lines that are too close together... If it is greater than the set value, and if it is too far away, it will be less than the set value;

[0019] Redundant GPS and altimeter / accelerometer, CCD industrial camera, and onboard host obtain precisely focused images. The onboard host analyzes continuous images captured by the CCD industrial camera. When the drone flies over geographic markers, inertial and GPS guidance ensures the captured images are in quasi-focus. GPS navigation error is 5-10 meters, requiring only 1-2 seconds to adjust to precise focus. During this process, the captured images are processed using Fourier optical transform. When identifying precise geographic markers, only the marked area needs to be processed, not the entire image frame, resulting in high processing speed. The Fourier optical imaging formula is:

[0020] (3)

[0021] The complex amplitude transmittance of the lens is Input various parameters to calculate the complex amplitude distribution of the image plane. When calculating near the real focal plane, refine the calculation and determine... and The numerical relationship between them is shown through two near-focus images. and The group determines the state of focus. .

[0022] In practical applications, two frames before and after the image is in focus are selected for fitting. The second moment of the frame before the image is in focus will continuously increase, while the second moment of the frame after the image is in focus will decrease. The reasons for selecting the image in focus are: (1) the objects are highly similar and can be directly compared linearly; (2) the focus is clear and has the highest signal-to-noise ratio compared with the image in defocus. The complex amplitude distribution of the image plane is calculated. When the calculation is near the real focus plane, the calculation can be densified to determine the image in focus. and The numerical relationship between them can be seen through two near-focus images. and The group determines the state of focus. .

[0023] Furthermore, in step 4, obtaining the optimal focused image or analyzing the image calculated according to Fourier optics, the camera coordinate system is used when solving for the camera target image. The camera is often in the world coordinate system, and the camera is treated as a rigid body, described using the Einstein convention of tensors. In the world coordinate system WGS84 and other Earth model coordinate systems, the basis vectors are: Assuming the camera's position Any target location is The basis vector of the camera coordinate system is , With fixed coordinate system The direction cosine of the included angle is The radius vector of the target point in the world coordinate system relative to the camera coordinate system is Coordinates in the camera coordinate system = , = = = . And add an inverter:

[0024] = (4)

[0025] After using an inverter, the x-component of the image point is obtained: .x= Similarly, take the y component .y= , These represent the physical widths of the CCD network in the X and Y directions, respectively.

[0026] The first and second terms are the rectangular x / y coordinate components of the geographic identifier and the camera optical center in the WGS84 coordinate system, respectively. In the iterative calculation, the latitude, longitude and elevation (φ, α, h) of the geographic identifier and the camera optical center are input for calculation.

[0027] Besides tensor algebra description, coordinate transformations of vectors (first-order tensors) are performed in matrix form for various types of vectors (such as position vectors, velocities, accelerations, etc.). The vector transformation form during coordinate transformation is as follows:

[0028]

[0029] In the above formula, This represents the vector in both the old and new coordinate systems. This is a transformation matrix, and its notation is the same as that of a tensor.

[0030] You need to input the direction cosine of the camera coordinate system basis vector with respect to the WGS84 coordinate system basis vector. During installation, after adding the inverter, make the direction of the camera coordinate system the same as the direction of the acceleration carrier coordinate system. The transformation from the carrier camera coordinate system to the WGS84 coordinate system is completed through two basis vector transformations: (1) The basis vector of the carrier coordinate system is transformed into the navigation coordinate system, and its transformation matrix is ​​the transpose matrix of the coordinate transformation. The form of the transformation matrix is ​​determined by the type of the navigation coordinate system; (2) the transformation matrix M84 of the basis vector from the navigation coordinate system to the WGS84 coordinate system; for example, the transformation matrix of NED (Northeast Earth) is selected. for

[0031]

[0032] (5)

[0033] The M84 projection matrix varies depending on the carrier coordinate system and the corresponding navigation coordinate system. Furthermore, the M84 projection matrix changes with the local geographic coordinate system and the form of the geodetic reference ellipsoid, such as for Northeast-Eastern Sky (ENU), where the corresponding projection matrix M84 = The total transformation matrix The choice of these reference frames will change.

[0034] During high-speed flight of a UAV on a predetermined or autonomous flight path, its side / below high-speed industrial camera rapidly captures single-lens images containing at least two precise marker images. Each image contains at least two pairs of u, v information—at least four. According to the lens imaging rules, the geographic marker image coordinates (u, v) on any image depend on the camera's latitude, longitude, and elevation position (φ, α, h), and three absolute attitude angles (φ, α, h). ); and the precision accelerometer has already accurately provided the pitch and roll angles ( ); Therefore, the four parameters of two geographic markers in an image can be used to calculate the position vector of the optical center, the three parameters of the radius vector, and the yaw angle to be determined. .

[0035] Obviously, in order to find the four unknowns of optical center location and orientation, four unrelated known reference quantities must be input into the solution model. Unrelated means that any one geographic landmark is not on the line connecting other geographic landmarks. Because camera mapping imaging is a linear mapping, if multiple geographic landmarks form a straight line, then the position vector of the landmarks is a linear combination of the first and second landmarks, and the mapping imaging cannot provide additional information.

[0036] Conversion between accelerometer and local Cartesian coordinate system: The accelerometer and CCD camera are mounted parallel and coaxially, and their body coordinate system basis vectors are in the same direction. The multi-element accelerometer is the core of the "flight control" and a key sensor for wind measurement, providing high-precision input parameters for UAV wind measurement models.

[0037] By transforming multiple coordinate systems and using the output of an accelerometer with three heading angles, a set of equations was established to invert the optical center position and attitude from precise geotagged images. Through iterative computer calculations, the spatial position and attitude information of the UAV represented by the optical center were accurately solved, and the UAV's speed and attitude angles were accurately determined.

[0038] The z-axis angle (i.e., heading angle) given by the accelerometer has low accuracy, typically 1. While it cannot meet the accuracy requirements, it can be used as the initial input for iterative calculations, such as expanding left and right by 1 degree Celsius from the current heading angle. Serving as an iterative calculation interval allows the calculation to converge as quickly as possible.

[0039] Furthermore, in step 5, based on the UAV's positions (φ1, α1, h1) and (φ2, α2, h2) consecutively represented by the optical center, with a time interval of dt, the UAV's speed is calculated: r= = [-a*sin( 1)cos(φ)i-bsin( sin(φ1)j+ ccos( )k] +[- a*cos( 1)sin(φ1)i+ bcos( The drone's climb speed is (h2-h1) / dt; the drone's eastward velocity is [- a*cos(α1)sin(φ1)i+ bcos(α1)cos(φ1)j]dφ / dt, and its northward velocity is [-a*sin( )cos(φ1)i- bsin( 1)sin(φ1)j+ ccos( 1)k] / dt, where dφ=φ2-φ1, dα=α2-α1.

[0040] Furthermore, in step 6, the speed of the ultrasonic anemometer is determined by angular velocity or directly by the change in position of the ultrasonic anemometer.

[0041] Furthermore, for accelerometers with angular velocity output, the speed of an ultrasonic anemometer can be calculated from the angular velocity. Since the angular velocity is always the same at any point on a rigid body, the relationship between the speed of the ultrasonic anemometer and the speed of the optical center is as follows: ; The angular velocity is given directly by the accelerometer in rough calculations. However, for precise calculations, the more accurate yaw angle derived from the camera mapping equation should be used instead of the accelerometer's output yaw angle. The variation in yaw angle can then be used to determine the yaw angle in that direction. The amount.

[0042] In the carrier coordinate system, the position of the ultrasonic anemometer is: This represents the position vector (Camera Center) of the center of the ventilation layer of the ultrasonic anemometer (UltraSonic or Pitot tube) relative to the optical center of the camera. In the body coordinate system, it is a constant vector determined during installation. The accelerometer provides values ​​in a body coordinate system. When using a body "front-right-lower" coordinate system... In equation (5) Other body coordinate systems and local coordinate systems also have corresponding transformation formulas. Among the three heading angles in the transformation matrix, the pitch and roll angles use the output of the accelerometer, while the heading angle is the result of iterative calculation, so that the data is more accurate and can match the output results of potentially higher-precision ultrasonic anemometers.

[0043] Furthermore, the speed can also be directly determined by the positional change of the ultrasonic anemometer. With the ultrasonic anemometer fixed in place, the position vector of the ultrasonic anemometer in the local coordinate system can be obtained using matrix transformation. ,in The position of the optical center in the local coordinate system is determined by optical imaging; the difference between the two values ​​divided by time dt is the ultrasonic anemometer speed. / dt. If a more accurate yaw angle is obtained by substituting it into the camera mapping equation, the result is... .

[0044] Furthermore, in step 7, knowing the ultrasonic anemometer's position and attitude, flight speed, and reading wind speed, information such as rotor speed (or propeller speed), acceleration and angular acceleration from the accelerometer, atmospheric temperature, and pressure can be read from the control system to calculate the actual wind speed from the dynamic equations. The narrow wind gap of the ultrasonic anemometer and the adjustment of the UAV's flight speed and heading are considered. Finally, software removes noise. The ultrasonic anemometer's readings can be considered unaffected by rain, snow, or other media. However, regardless of whether it's a propeller, rotor, or other power source, the UAV's flight against gravity will inevitably disturb the airflow around the ultrasonic anemometer. When the ultrasonic anemometer's flight speed deviates from the horizontal direction, the top and bottom of the anemometer will not interfere with the measurement. This interference can be approximated through numerical simulation or wind tunnel experiments. PID control methods can be used to continuously optimize the "flight control" module using the accelerometer output, ensuring that the ultrasonic sensor's flight speed is approximately horizontal or even in near-parallel flight. Only in parallel motion... , .

[0045] Based on the principle of relativity, wind speed V w Ultrasonic speed measurement ultrasonic relative air velocity Among the three = -V w Numerical simulations or wind tunnel experiments can be used to approximate the measurement and correct for excessive deviations in the ultrasonic anemometer during horizontal flight. By following the wind, the error caused by the drone's flight can be greatly reduced, and may even be significantly lower than the measurement error when the drone is fixed in place.

[0046] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: the spatial positioning of the present invention is accurate to the cm level and the time resolution is accurate to 1ms, with both spatiotemporal resolution improved by more than two orders of magnitude; during measurement, the flight control system allows the UAV to "follow the wind" as much as possible, that is, the relative air speed (ultrasonic anemometer reading) on ​​the UAV approaches 0, at which point the wind speed is equal to the UAV's flight speed, and the measurement error approaches 0. This not only eliminates the additional error caused by flight speed, but can even be much smaller than the fixed wind measurement error, which has high engineering application value. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the system structure of the present invention.

[0048] Figure 2 This is a schematic diagram of the method flow of the present invention.

[0049] Figure 3 This is a schematic diagram of various coordinate systems involved in the present invention.

[0050] Figure 4 This is a schematic diagram illustrating the decomposition and transformation relationship of basis vectors between different coordinate systems in this invention. Detailed Implementation

[0051] like Figure 1 As shown, a UAV wind measurement system includes: an ultrasonic anemometer, redundant GPS and altimeter, a high-precision accelerometer with three heading angle accelerometers, a CCD industrial camera, a camera optical center, high-precision geotags, a ground host, an airborne host, and a UAV platform. The system inputs a high-precision geotag database. The UAV platform carries the airborne host, ultrasonic anemometer, redundant GPS and altimeter, high-precision accelerometer with three heading angle accelerometers, and CCD industrial camera. The UAV climbs to a height above the geotags and takes a picture using the CCD industrial camera, capturing geotags with at least two sufficiently spaced geotags. The UAV flies as horizontally as possible. The system iteratively calculates the current optical center position of the UAV, the UAV's speed, and the speed of the ultrasonic anemometer. Finally, it calculates the wind speed to reduce measurement errors.

[0052] like Figures 2-4 As shown, a method for wind measurement using a drone includes the following steps:

[0053] Step 1: Input the high-precision geographic identification database into the system;

[0054] Step 2: The UAV platform is equipped with an onboard host, an ultrasonic anemometer, redundant GPS and altitude accelerometer, a high-precision accelerometer with 3 heading angle accelerometers and a CCD industrial camera;

[0055] Step 3: The drone climbs and flies to above the sign, and takes a picture with the CCD industrial camera above the sign. The drone should fly as horizontally as possible.

[0056] Step 4: Iteratively solve for the current optical center position of the UAV;

[0057] Step 5: Determine the speed of the UAV;

[0058] Step 6: Calculate the speed measured by the ultrasonic anemometer;

[0059] Step 7: Calculate the wind speed to reduce measurement error.

[0060] In step 1, the system inputs a high-precision geoidentification database, including GPS information and image information, which are two-dimensional or three-dimensional. The high-precision geoidentification database is stored simultaneously in the ground host and the airborne host.

[0061] In step 3, the drone climbs and flies above the marker, where a CCD industrial camera takes a picture. The drone should fly as horizontally as possible to ensure the image is in focus.

[0062] (1)

[0063] (2)

[0064] Where f is the focal length, u is the object distance, and v is the phase distance. For a fixed-focus f-lens and a fixed distance v, the object distance u for precise focusing and the magnification are... It is definite; in the camera coordinate system, the optical center is the origin, the CCD chip's X-axis is the X-axis, and the Z-axis is outward along the optical axis. The lens mapping law is written as... =- = , among which the target location Let be the position vector of the target geographic marker relative to the optical center; the CCD industrial camera continuously captures continuous images of a relatively long line connecting the current geographic marker at high speed, and obtains the optimal focused image according to the Fourier optical imaging transform and the magnification of the line connecting the two markers; within the depth of field, lines that are too close together... If it is greater than the set value, and if it is too far away, it will be less than the set value;

[0065] Redundant GPS and altimeter / accelerometer, CCD industrial camera, and onboard host obtain precisely focused images. The onboard host analyzes continuous images captured by the CCD industrial camera. When the drone flies over geographic markers, inertial and GPS guidance ensures the captured images are in quasi-focus. GPS navigation error is 5-10 meters, requiring only 1-2 seconds to adjust to precise focus. During this process, the captured images are processed using Fourier optical transform. When identifying precise geographic markers, only the marked area needs to be processed, not the entire image frame, resulting in high processing speed. The Fourier optical imaging formula is:

[0066] (3)

[0067] The complex amplitude transmittance of the lens is Input various parameters to calculate the complex amplitude distribution of the image plane. When calculating near the real focal plane, refine the calculation and determine... and The numerical relationship between them is shown through two near-focus images. and The group determines the state of focus. .

[0068] In practical applications, two frames before and after the image is in focus are selected for fitting. The second moment of the frame before the image is in focus will continuously increase, while the second moment of the frame after the image is in focus will decrease. The reasons for selecting the image in focus are: (1) the objects are highly similar and can be directly compared linearly; (2) the focus is clear and has the highest signal-to-noise ratio compared with the image in defocus. The complex amplitude distribution of the image plane is calculated. When the calculation is near the real focus plane, the calculation can be densified to determine the image in focus. and The numerical relationship between them can be seen through two near-focus images. and The group determines the state of focus. .

[0069] In step 4, the optimal focused image or the image calculated according to Fourier optics is obtained. When solving for the camera target image, the camera coordinate system is used. The camera is often in the world coordinate system, and is considered a rigid body, described using the Einstein convention of tensors. In the world coordinate system WGS84 and other Earth model coordinate systems, the basis vectors are: Assuming the camera's position Any target location is The basis vector of the camera coordinate system is , With fixed coordinate system The direction cosine of the included angle is The radius vector of the target point in the world coordinate system relative to the camera coordinate system is Coordinates in the camera coordinate system = , = = = . And add an inverter:

[0070] = (4)

[0071] Take the x-component of the image point: .x= Similarly, take the y component .y= , .

[0072] The first and second terms are the rectangular x / y coordinate components of the geographic identifier and the camera optical center in the WGS84 coordinate system, respectively. In the iterative calculation, the latitude, longitude and elevation (φ, α, h) of the geographic identifier and the camera optical center are input for calculation.

[0073] You need to input the direction cosine of the camera coordinate system basis vector with respect to the WGS84 coordinate system basis vector. During installation, after adding the inverter, ensure the camera coordinate system is aligned with the accelerometer coordinate system. This is achieved through two basis vector transformations: Step 1 – Transform the base vectors of the carrier (camera) coordinate system into the navigation coordinate system; the transformation matrix is ​​the transpose of the coordinate transformation. The form of the transformation matrix is ​​determined by the type of navigation coordinate system. In the NED (Northeast) navigation coordinate system, the projection matrix M84 from any NED local coordinate system basis vector to the WGS84 coordinate system basis vector is... for

[0074] =M84 =

[0075] (5)

[0077] The semi-major axis of the WGS84 geodetic ellipsoid is a=b=6378137.0000 meters, and the minor axis is c=6356752.31414 meters; These are yaw angle, pitch angle, and roll angle, respectively.

[0078] During high-speed flight of a UAV on a predetermined or autonomous flight path, its side / below high-speed industrial camera rapidly captures images containing two or more precise markers in a single lens imaging process. Each image contains at least two pairs of u, v information—at least four. According to the lens imaging rules, the coordinates (u, v) of the geographic marker image in any image depend on the camera position (φ, α, h) and the three absolute attitude angles (φ, α, h). ); and the precision accelerometer has already provided accurate ( Therefore, the four parameters of two geographic markers in an image can be used to calculate the position vector of the optical center, the three parameters of the radius vector, and the yaw angle to be determined. .

[0079] Obviously, in order to find the four unknowns of optical center location and orientation, four unrelated known reference quantities must be input into the solution model. Unrelated means that any one geographic landmark is not on the line connecting other geographic landmarks. Because camera mapping imaging is a linear mapping, if multiple geographic landmarks form a straight line, then the position vector of the landmarks is a linear combination of the first and second landmarks, and the mapping imaging cannot provide additional information.

[0080] Conversion between accelerometer and local Cartesian coordinate system: The accelerometer and CCD camera are mounted parallel and coaxially, and their body coordinate system basis vectors are in the same direction. The multi-element accelerometer is the core of the "flight control" and a key sensor for wind measurement, providing high-precision input parameters for UAV wind measurement models.

[0081] By transforming multiple coordinate systems and using the output of an accelerometer with three heading angles, a set of equations was established to invert the optical center position and attitude from precise geotagged images. Through iterative computer calculations, the spatial position and attitude information of the UAV represented by the optical center were accurately solved, and the UAV's speed and attitude angles were accurately determined.

[0082] The z-axis angle (i.e., heading angle) given by the accelerometer has low accuracy, typically 1. While it cannot meet the accuracy requirements, it can be used as the initial input for iterative calculations, such as expanding left and right by 1 degree Celsius from the current heading angle. Serving as an iterative calculation interval allows the calculation to converge as quickly as possible.

[0083] In step 5, based on the UAV's positions (φ1, α1, h1) and (φ2, α2, h2) consecutively represented by the optical center, with a time interval of dt, the UAV's speed is calculated: r= = [-a*sin( 1)cos(φ)i- bsin( sin(φ1)j+ ccos( )k] +[- a*cos( 1)sin(φ1)i+ bcos( The drone's climb speed is (h2-h1) / dt; the drone's eastward velocity is [- a*cos(α1)sin(φ1)i+ bcos(α1)cos(φ1)j]dφ / dt, and its northward velocity is [-a*sin( )cos(φ1)i- bsin( 1)sin(φ1)j+ ccos( 1)k] / dt. Among them, dφ=φ2-φ1, dα=α2-α1.

[0084] In step 6, the speed of the ultrasonic anemometer is determined by angular velocity or directly by the change in position of the ultrasonic anemometer.

[0085] The ultrasonic anemometer speed is calculated by using angular velocity. In conditions such as strong winds and rapid currents, and during pitch and roll maneuvers of a UAV, the ultrasonic anemometer speed is determined because the angular velocity at any point on a rigid body is always the same. With the speed of light The relationship is as follows: ; It is the transformation matrix from the vehicle coordinate system to the navigation coordinate system. The pitch and roll angles in the transformation matrix use the output of the accelerometer, while the yaw angle is calculated iteratively. This makes the data more accurate, allowing it to match the potentially higher-precision output of the ultrasonic anemometer. The angular velocity of the UAV in the carrier coordinate system can be calculated from the changes in three precise attitude angles. The angular velocity components corresponding to the pitch and roll angles can be directly obtained from the output values ​​of the accelerometer. This represents the position vector (Camera Center) of the center of the ventilation layer of the ultrasonic anemometer (UltraSonic or Pitot tube) relative to the optical center of the camera. In the body coordinate system, it is a constant vector determined during installation.

[0086] The speed can also be directly determined by the positional change of the ultrasonic anemometer. The ultrasonic anemometer is fixedly installed; in the carrier coordinate system, the position of the ultrasonic anemometer is... The position vector of the ultrasonic anemometer in the local coordinate system is obtained based on the matrix transformation formula. ,in The position of the optical center in the local coordinate system is determined by optical imaging; the difference between the two values ​​divided by time dt is the ultrasonic anemometer speed. / dt.

[0087] In step 7, knowing the ultrasonic anemometer's position and attitude, flight speed, and reading wind speed, information such as rotor speed (or propeller speed), acceleration and angular acceleration from the accelerometer, atmospheric temperature, and pressure can be read from the control system to calculate the actual wind speed from the dynamic equations. The narrow wind tunnel of the ultrasonic anemometer and the adjustment of the UAV's flight speed and heading are considered. Finally, software removes noise. The ultrasonic anemometer's readings can be considered unaffected by rain, snow, or other media. However, regardless of whether it's a propeller, rotor, or other power source, the UAV's flight against gravity will inevitably disturb the airflow around the ultrasonic anemometer. When the ultrasonic anemometer's flight speed deviates from the horizontal direction, the top and bottom of the anemometer will not interfere with the measurement. This interference can be approximated through numerical simulation or wind tunnel experiments. PID control methods can be used to continuously optimize the "flight control" module using the accelerometer output, ensuring that the ultrasonic sensor's flight speed is approximately horizontal or even in near-parallel flight. Only in parallel motion (i.e.,...) will the wind speed be affected. . .

[0088] Based on the principle of relativity, wind speed V w Ultrasonic speed measurement ultrasonic relative air velocity Among the three = -V w .

[0089] Example 1: The UAV is equipped with a high-precision accelerometer with three heading angles (acceleration, angle, and angular velocity), and redundantly equipped with a general 10-element ordinary accelerometer (including a constant altitude barometer) and an accelerometer. The output uses the Northeastern Universe (ENU) navigation system, with the carrier reference frame being "Right Front Upper" (RFU). The transformation matrix from the carrier coordinate system to the navigation coordinate system is also included. The UAV platform is a vertically takeoff fixed-wing aircraft with a USB 3.0 interface, requiring no additional power cable. The accompanying industrial lens has a distortion of less than 0.1%, and after calibration, it can reach the level of 0.01%. The optical system meets the centimeter-level resolution requirements of the UAV at an altitude of 200 meters.

[0090] The drone is a fixed-wing drone that can take off and land vertically. It has a wingspan of 2400 mm and can fly continuously for 3 hours in a level 4 wind when carrying a 1 kg payload.

[0091] Accelerometer and optical system provide accuracy of over 0.1%. Typical positioning error for UAVs at a height of 100 meters is 2-5 cm. The microcomputer uses a Unix-like system with a response time of 1 ms.

[0092] Example 2: Positioning using BeiDou signals, employing the 2000 China Geodetic Coordinate System (CGCS2000), a high-precision 6-element accelerometer (accelerometer and angle), with the accelerometer outputting a coordinate transformation matrix from the carrier coordinate system to the navigation coordinate system using Northeast-East (NED) navigation and the carrier reference frame being "Front Right-Down" (FRD); a redundant ordinary 10-element accelerometer; and a binocular system composed of two 1-inch 8.8-megapixel industrial cameras. This binocular system can quickly perceive the target's z-direction value. Compared to the 2.3-megapixel camera in Example 1, the number of ground geographic markers can be reduced to 1 / 4, or the camera positioning accuracy can be doubled. A liquid zoom lens is used, with a constant distance u. The liquid zoom lens's electronic control hardware and software can continuously and rapidly change the lens focal length f, performing Fourier transforms on continuous images to ensure precise focusing. By analyzing the distance between the two geographic marker image points, the magnification factor can be calculated. Substituting directly into equation (4), since the stability of the coefficient matrix calibrated by the camera is lower than that of a solid fixed-focus lens, its accuracy is slightly lower than that of a fixed-focus lens, but still much higher than that of the traditional GPS positioning wind measurement method, and it has the advantage of rapid measurement of the entire elevation. The major axis of WGS84 is 6378137.0000 meters, and the minor axis is 6356752.31414 meters. Compared with WGS84, the minor axis of the Beidou reference ellipsoid differs by 0.06 mm. Beidou can also output latitude, longitude and elevation information, so it is highly compatible with WGS84. The aforementioned processing theories and methods are exactly the same, and the UAV is a rotorcraft.

Claims

1. An unmanned aerial vehicle wind measurement system, characterized in that, It includes: Ultrasonic anemometer, redundant GPS and height accelerometer, high-precision 3 heading angle accelerometer, CCD industrial camera, camera optical center, high-precision geographic location, ground host, airborne host and unmanned aerial vehicle platform; the system inputs high-precision geographic location database, the unmanned aerial vehicle platform is equipped with airborne host, ultrasonic anemometer, redundant GPS and height accelerometer, high-precision 3 heading angle accelerometer and CCD industrial camera; the unmanned aerial vehicle climbs to the top of the identification, takes a picture above the identification with the CCD industrial camera, and shoots a geographic location with more than two long intervals; the unmanned aerial vehicle flies as horizontally as possible; the current optical center position of the unmanned aerial vehicle is iteratively solved, the speed of the unmanned aerial vehicle is solved, and the speed of the ultrasonic anemometer is solved; the wind speed is solved, and the measurement error is reduced by using wind chasing navigation.

2. A method for measuring wind based on the unmanned aerial vehicle wind measurement system according to claim 1, characterized in that, It includes the following steps: Step 1, input high-precision geographic location database in the system; Step 2, the unmanned aerial vehicle platform is equipped with airborne host, ultrasonic anemometer, redundant GPS and height accelerometer, high-precision 3 heading angle accelerometer and CCD industrial camera; Step 3, the unmanned aerial vehicle climbs to the top of the identification near the specified height, takes a picture above the identification with the CCD industrial camera, and shoots a geographic location with more than two long intervals; the unmanned aerial vehicle flies as horizontally as possible; Step 4, iteratively solve the current optical center position of the unmanned aerial vehicle; Step 5, solve the speed of the unmanned aerial vehicle; Step 6, solve the speed of the ultrasonic anemometer; Step 7, solve the wind speed and reduce the measurement error. 3.The UAV wind measurement method of claim 2, wherein, In step 1, the system inputs high-precision geographic location database, including GPS information and image information, which is two-dimensional or three-dimensional, and the high-precision geographic location database is stored in the ground host and the airborne host at the same time. 4.The UAV wind measurement method of claim 2, wherein, In step 3, the unmanned aerial vehicle climbs to the top of the identification, takes a picture above the identification with the CCD industrial camera, and the unmanned aerial vehicle flies as horizontally as possible; the focused image satisfies （1） （2） Where f is the focal length, u is the object distance, and v is the phase distance. For a fixed-focus f-lens and a fixed distance v, the object distance u for precise focusing and the magnification are... It is definite; in the camera coordinate system, the optical center is the origin, the X-axis of the CCD industrial camera is the X-axis, and the Z-axis is outward along the optical axis. The lens mapping law is written as... =- = , among which the target location Let be the position vector of the target geographic marker relative to the optical center; the CCD industrial camera continuously captures continuous images of a relatively long line connecting the current geographic marker at high speed, and obtains the optimal focused image according to the Fourier optical imaging transform and the magnification of the line connecting the two markers; within the depth of field, lines that are too close together... If it is greater than the set value, and if it is too far away, it will be less than the set value; The redundant GPS and height accelerometer, CCD industrial camera and airborne host obtain accurate focused images, the airborne host analyzes the continuous images taken by the CCD industrial camera, when the unmanned aerial vehicle sweeps through the geographic location, the Fourier optical transform is processed to take images, the Fourier optical imaging formula is: （3） The complex amplitude transmittance of the lens is , inputting various parameters, calculating the complex amplitude distribution of the image surface, when calculating the vicinity of the real focal plane, determining the numerical relationship between and , determining the real focal state of by two near real focal state and groups. 5.The UAV wind measurement method of claim 2, wherein, In step 4, the best focus image or image analysis according to Fourier optical calculation is obtained, and the camera coordinate system is used in solving the camera target image. The camera is regarded as a rigid body, described by the Einstein convention of tensor, and the world coordinate system WGS84 earth model coordinate system is as follows: each base vector is , the position of the camera is assumed to be , the position of any target is , the base vector of the camera coordinate system is , , the direction cosine of the angle between the fixed coordinate system is , the vector of the target point to the camera coordinate system in the world coordinate system is , the coordinate in the camera coordinate system is = , = = = . ; And add an inverter: = （4） After the inverter, take the image point x component: .x= , and the y component .y= , respectively as the CCD network X / Y direction physical width; ) the former and the latter are respectively the geographic identification of WGS84 coordinate system and the rectangular x / y coordinate components of the camera optical center, in the iterative calculation, the longitude, latitude and elevation (φ, α, h) of the geographic identification and the camera optical center are calculated respectively; After adding the inverter, let the camera coordinate system direction be the same as the acceleration carrier coordinate system direction, and complete the carrier camera coordinate system to WGS84 coordinate system transformation through two times of base vector transformation: (1) base vector transformation of the carrier coordinate system to the navigation coordinate system, and the conversion matrix is the transpose matrix of the coordinate transformation , and the form of the conversion matrix is determined by the type of the navigation coordinate system; (2) the conversion matrix M84 of the base vector from the navigation coordinate system to the WGS84 coordinate system; and the total conversion matrix of the NED (north, east, and ground) =M84 = （5） M84 varies with the carrier coordinate system and the corresponding navigation coordinate system, while M84 also varies with the form of the navigation local geographic coordinate system and the geodetic reference ellipsoid. The total transformation matrix The choice of these factors can vary; The high-speed industrial camera on the side / below the UAV takes a single-lens image containing more than two precise landmark images during high-speed flight on a predetermined or autonomous flight path. One image contains at least two pairs of u, v information - at least four. According to the lens imaging law, the geographic landmark image coordinates (u, v) on any one image depend on the camera latitude and longitude and elevation position (φ, α, h), three absolute attitude angles (θ, φ, ψ) ); and the precise acceleration sensor has accurately given the pitch angle and roll angle (θ, φ) ), so the four parameters of the two geographic landmarks in one image can be used to calculate the position vector of the optical center. The three parameters and the yaw angle to be solved . 6.The UAV wind measurement method of claim 2, wherein, In step 5, based on the UAV's positions (φ1, α1, h1) and (φ2, α2, h2) consecutively represented by the optical center, with a time interval of dt, the UAV's speed is calculated: r= = [-a*sin( 1)cos(φ)i- bsin( sin(φ1)j+ ccos( )k] +[- a*cos( 1)sin(φ1)i+ bcos( The drone's climb speed is (h2-h1) / dt; the drone's eastward velocity is [- a*cos(α1)sin(φ1)i+ bcos(α1)cos(φ1)j]dφ / dt, and its northward velocity is [-a*sin( cos(φ1)i- bsin( 1)sin(φ1)j+ ccos( 1)k] / dt, where dφ=φ2-φ1, dα=α2-α1. 7.The UAV wind measurement method of claim 2, wherein, In step 6, the speed of the ultrasonic anemometer is solved by angular velocity or directly solved by the position change of the ultrasonic anemometer. 8.The UAV wind measurement method of claim 7, wherein, For the accelerometer with angular velocity output, the ultrasonic anemometer speed is solved by angular velocity. The angular velocity of any point of rigid body is always equal. The relationship between the ultrasonic anemometer speed and the optical center speed is as follows: ; represents the angular velocity. In rough calculation, it is directly given by the acceleration sensor. In accurate calculation, the more accurate yaw angle solved by the camera mapping equation should be used to replace the output yaw angle of the acceleration sensor, and the component of the direction is solved by the change. 9.The UAV wind measurement method of claim 7, wherein, The position and attitude angle of the ultrasonic anemometer are directly used to obtain the navigation speed of the ultrasonic anemometer, the ultrasonic anemometer is fixedly installed, the position of the ultrasonic anemometer in a carrier coordinate system is ; the position vector of the ultrasonic anemometer in a local coordinate system is obtained according to a matrix transformation formula , wherein is the position of the optical center in the local coordinate system obtained by an optical imaging method; the difference between the front and rear two times is divided by the time dt, that is, the navigation speed of the ultrasonic anemometer / dt. 10.The UAV wind measurement method of claim 2, wherein, In step 7, knowing the position and attitude of the ultrasonic anemometer, the speed and the reading wind speed, from the control system to read the rotor speed, acceleration sensor acceleration and angular acceleration, atmospheric temperature and pressure information from the dynamics equation to calculate the actual wind speed, the narrow wind layer of the ultrasonic anemometer has been removed. The speed and direction of the unmanned aerial vehicle are adjusted, and finally the noise is removed. The PID control method is adopted, the output of the acceleration sensor is constantly optimized, and the ultrasonic sensor speed can be ensured to be approximately horizontal or even in horizontal flight and close to the flat motion. Only in the parallel motion, that is or ; According to the relativity principle, the wind speed V w , the ultrasonic meter speed , the ultrasonic meter relative air speed between = -V w .

Images