A vertical profile observation method and device based on a UAV

Abstract

This invention discloses a method and apparatus for vertical profile observation based on unmanned aerial vehicles (UAVs), relating to the field of UAV environmental detection technology. The method collects high-frequency inertial vibration and electrically regulated duty cycle sequences of the aircraft during its dwell phase, inverts the rotor thrust distribution, and calibrates the wake envelope. It extracts three-axis attitude deflection angles to invert the environmental wind field vector, spatially superimposes it with the envelope surface to delineate the windward, undisturbed sampling area, generates yaw-compensated waypoints for controlled climb, and simultaneously collects data. It performs time-domain alignment on the observation and trajectory sequences, dynamically integrates the sensor response time constant with the airflow disturbance rate to generate altitude hysteresis compensation parameters, and injects these parameters back into the trajectory to reconstruct the hysteresis-corrected altitude sequence. It extracts gradient vectors through vertical difference, retains the original sampling density in extreme value abrupt change intervals, performs feature aggregation and dimensionality reduction in smooth intervals, and finally outputs a non-equidistant atmospheric vertical profile sequence. The apparatus described above is used to implement the above method.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) environmental detection technology, specifically to a method and apparatus for vertical profile observation based on UAVs. Background Technology

[0002] Vertical profile observations of atmospheric boundary layer physical and chemical parameters play a fundamental role in weather forecasting, environmental pollution diffusion analysis, and disaster prevention, mitigation, and early warning. With the opening of low-altitude airspace and the miniaturization of intelligent sensing devices, vertical sounding observations using multi-rotor UAVs equipped with environmental sensors are gradually becoming an important auxiliary means of modern atmospheric sounding, thanks to their flexible networking and high vertical spatial coverage. This type of approach typically involves pre-setting the UAV's vertical ascent trajectory and simultaneously recording sensor data to obtain the distribution patterns of atmospheric elements on a spatial scale.

[0003] However, in actual observation operations, multi-rotor aircraft inevitably generate strong rotor downwash, and the aircraft must frequently adjust its attitude to counteract environmental wind fields. Existing technologies often employ a direct binding logic between positioning altitude and instantaneous observation data processing. This fixed-mount and spatially synchronized mapping method is insufficiently adaptable to the distortions and disturbances in the microscale flow field where the meteorological probe is located. The local native atmosphere at the sensor inlet is prematurely agitated and heated by the rotor wake. Simultaneously, the inherent hardware response delay of the sensor can lead to severe measurement altitude misalignment during high-speed climbs, easily resulting in problems such as smoothing of abrupt stratification features, positioning deviations at inversion layers or pollution boundaries, and distortion of observation data under different environmental wind speeds. Therefore, it is still necessary to provide a UAV-based vertical profile observation method to improve the accuracy of undisturbed flow field conditions and spatial matching during microscale meteorological observations. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a method and apparatus for vertical profile observation based on unmanned aerial vehicles (UAVs) to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a vertical profile observation method based on an unmanned aerial vehicle (UAV), comprising the following steps: S1, collecting the high-frequency inertial vibration sequence and electronically controlled duty cycle sequence of the UAV during its dwell phase, extracting the dominant frequency amplitude vector and combining it with the electronically controlled duty cycle to invert the instantaneous thrust distribution of the rotor, and calibrating the rotor wake envelope surface under the dynamic flow field; S2, extracting the three-axis attitude deflection sequence during the dwell phase to invert the environmental wind field vector, performing spatial vector superposition of the environmental wind field vector and the rotor wake envelope surface, delineating the windward undisturbed sampling area on the outer edge of the rotor wake envelope surface; tracking the azimuth plane of the windward undisturbed sampling area to generate a yaw compensation waypoint sequence, controlling the UAV to climb along the yaw compensation waypoint sequence, and simultaneously collecting atmospheric data. S3. Perform time-domain alignment on the atmospheric observation sequence and the 3D trajectory sequence, extract the sensor's response time constant, and invert the airflow disturbance rate by combining the rotor's instantaneous thrust distribution in the windward, undisturbed sampling area; dynamically integrate the response time constant and the airflow disturbance rate to generate an altitude hysteresis compensation parameter, and inject it back into the vertical axis of the 3D trajectory sequence to reconstruct the hysteresis-corrected altitude sequence; S4. Perform vertical difference on the atmospheric observation sequence that matches the hysteresis-corrected altitude sequence to extract the vertical gradient vector characterizing the stratification features; retain the original sampling density of meteorological nodes in the extreme value abrupt interval of the vertical gradient vector, and perform feature aggregation and dimensionality reduction on adjacent nodes in the smooth interval to output a non-equidistant atmospheric vertical profile sequence.

[0006] In a preferred embodiment, the specific process of collecting the high-frequency inertial vibration sequence and electronically controlled duty cycle sequence of the UAV during its dwell phase, extracting the dominant frequency amplitude vector, and combining it with the electronically controlled duty cycle to invert the instantaneous rotor thrust distribution, and calibrating the rotor wake envelope surface under the dynamic flow field is as follows: Performing a fast Fourier transform on the high-frequency inertial vibration sequence of the UAV to extract the energy concentration frequency band, and converting the energy concentration frequency band into the rotor rotation speed frequency; fusing the rotor rotation speed frequency and the electronically controlled duty cycle to solve the instantaneous rotor thrust distribution using the lift characteristic matrix; establishing the downward exhaust velocity gradient and turbulent dissipation rate distribution boundary based on the instantaneous rotor thrust distribution, mapping the distribution boundary to the body coordinate system, and calibrating the rotor wake envelope surface under the dynamic flow field.

[0007] In a preferred embodiment, the process of extracting the environmental wind field vector from the three-axis attitude deflection sequence during the dwell phase, superimposing the environmental wind field vector with the rotor wake envelope surface using spatial vector superposition, and delineating the windward undisturbed sampling area on the outer edge of the rotor wake envelope surface is as follows: decompose the three-axis attitude deflection sequence to extract pitch and roll components, and convert the pitch and roll components into horizontal drag compensation vectors according to the hovering force balance equation; fuse the horizontal drag compensation vector with the rotor instantaneous thrust distribution to invert the environmental wind field vector; extract the offset projection boundary of the environmental wind field vector onto the rotor wake envelope surface, and intercept the feature space unaffected by downwash flow shear at the windward outer edge of the offset projection boundary to delineate the windward undisturbed sampling area.

[0008] In a preferred embodiment, the specific process of generating a yaw-compensated waypoint sequence by tracking the azimuth plane of the windward undisturbed sampling area, controlling the UAV to climb along the yaw-compensated waypoint sequence, and simultaneously acquiring atmospheric observation sequences and three-dimensional trajectory sequences is as follows: Calculate the dynamic center coordinates of the windward undisturbed sampling area on the vertical trajectory; establish yaw constraint parameters by associating the dynamic center coordinates with the angle between the incoming environmental wind field vector and the incoming flow, generating a yaw-compensated waypoint sequence containing four dimensions: longitude, latitude, altitude, and yaw angle; drive the UAV to execute the yaw-compensated waypoint sequence, triggering the hardware-level clock synchronization latch signal between the atmospheric observation sensor acquisition end and the positioning module output end during the climb process, and acquiring atmospheric observation sequences and three-dimensional trajectory sequences.

[0009] In a preferred embodiment, the specific process of performing time-domain alignment on the atmospheric observation sequence and the three-dimensional trajectory sequence, extracting the sensor's response time constant, and inverting the airflow disturbance rate by combining the rotor instantaneous thrust distribution in the windward undisturbed sampling area is as follows: extract the hardware clock markers of the atmospheric observation sequence and the three-dimensional trajectory sequence, evaluate the clock drift offset between the hardware clock markers through the cross-correlation function, perform phase compensation resampling to output time-domain aligned observation trajectory synchronization data; analyze the probe thermal constant and gas diffusion flux of the environmental sensor and combine them with the current initial state of the environment to extract the response time constant of the nonlinear response interval; map the rotor instantaneous thrust distribution to the boundary of the windward undisturbed sampling area, calculate the flow field kinetic energy attenuation coefficient at the boundary, solve the forced convection acceleration component of the ambient atmosphere based on the flow field kinetic energy attenuation coefficient, and invert to generate the airflow disturbance rate.

[0010] In a preferred embodiment, the specific process of generating a height hysteresis compensation parameter by dynamically integrating the response time constant and the airflow disturbance rate, and then back-injecting it into the vertical axis of the three-dimensional trajectory sequence to reconstruct the hysteresis-corrected height sequence is as follows: A differential equation containing the convection terms of the response time constant and the airflow disturbance rate is constructed; numerical continuous integration is performed on the differential equation along the time axis to discretize and output the height hysteresis compensation parameter; the original height coordinate nodes on the vertical axis of the three-dimensional trajectory sequence are extracted; the height hysteresis compensation parameter is converted into a negative offset coordinate vector; the negative offset coordinate vector is driven to perform a spatial inverse topological substitution with the original height coordinate nodes; the topological substitution result is mapped to the original height coordinates to reconstruct the hysteresis-corrected height sequence.

[0011] In a preferred embodiment, the specific process of performing vertical differencing on the atmospheric observation sequence matching the hysteresis-corrected altitude sequence to extract the vertical gradient vector characterizing the stratification features is as follows: The hysteresis-corrected altitude sequence and the atmospheric observation sequence are subjected to spatial coordinate mapping to construct a spatial altitude-meteorological element correlation matrix; an adaptive variable step-size sliding window is configured along the spatial altitude dimension on the spatial altitude-meteorological element correlation matrix; high-order forward differencing is performed on the meteorological elements within the adaptive variable step-size sliding window to extract the slope fluctuation envelope of the meteorological elements as a function of altitude; the slope fluctuation envelope is converted into a vertical gradient vector containing the stratification temperature rate sequence and the substrate concentration jump rate sequence.

[0012] In a preferred embodiment, the specific process of retaining the original sampling density of meteorological nodes in the extreme value abrupt change interval of the vertical gradient vector, and performing feature aggregation and dimensionality reduction on adjacent nodes in the smooth interval to output a non-equidistant atmospheric vertical profile sequence is as follows: Calculate the first derivative of the vertical gradient vector, locate the abrupt boundary layer region where the absolute value of the derivative exceeds the limit and the uniform mixing layer region where the derivative approaches zero, label the abrupt boundary layer region as the extreme value abrupt change interval, and label the uniform mixing layer region as the smooth interval; retain the original sampling density of meteorological nodes within the extreme value abrupt change interval; extract the sequence of adjacent meteorological nodes within the smooth interval, configure an adaptive spatial distance clustering algorithm to calculate the physical feature centroid of adjacent node clusters, and extract the feature centroid as the representative node for dimensionality reduction; splice the meteorological nodes and the representative node for dimensionality reduction within the extreme value abrupt change interval, and reorder them according to the spatial height index to generate a non-equidistant atmospheric vertical profile sequence.

[0013] A UAV-based vertical profile observation device, used to execute the aforementioned UAV-based vertical profile observation method, includes: a wake calibration unit, used to collect the UAV's high-frequency inertial vibration sequence and electronically controlled duty cycle sequence during the UAV's dwell phase, extract the dominant frequency amplitude vector, and combine it with the electronically controlled duty cycle to invert the instantaneous rotor thrust distribution, calibrating the rotor wake envelope surface under dynamic flow field; a trajectory acquisition unit, used to extract the three-axis attitude deflection angle sequence during the dwell phase to invert the environmental wind field vector, perform spatial vector superposition of the environmental wind field vector and the rotor wake envelope surface, delineate the windward undisturbed sampling area on the outer edge of the rotor wake envelope surface; track the azimuth plane of the windward undisturbed sampling area to generate a yaw compensation waypoint sequence, control the UAV to climb along the yaw compensation waypoint sequence, and simultaneously collect atmospheric data. The system consists of an observation sequence and a 3D trajectory sequence; a hysteresis correction unit, used to perform temporal alignment between the atmospheric observation sequence and the 3D trajectory sequence, extract the sensor's response time constant, and invert the airflow disturbance rate by combining the rotor's instantaneous thrust distribution in the windward, undisturbed sampling area; dynamically integrating the response time constant and the airflow disturbance rate to generate a height hysteresis compensation parameter, which is then back-injected into the vertical axis of the 3D trajectory sequence to reconstruct the hysteresis-corrected height sequence; and a profile generation unit, used to perform vertical difference on the atmospheric observation sequence that matches the hysteresis-corrected height sequence, extracting the vertical gradient vector characterizing stratification features; preserving the original sampling density of meteorological nodes in the extreme value abrupt intervals of the vertical gradient vector, and performing feature aggregation and dimensionality reduction on adjacent nodes in the smooth intervals, outputting a non-equidistant atmospheric vertical profile sequence.

[0014] The technical effects and advantages of this invention are as follows: (1) A vertical profile observation method based on UAVs involves collecting high-frequency inertial vibration sequences and electronically controlled duty cycle sequences of the UAV during its dwell phase, inverting the instantaneous thrust distribution of the rotor, calibrating the rotor wake envelope under the dynamic flow field, extracting the three-axis attitude deflection sequence to invert the environmental wind field vector, and superimposing it with the spatial vector of the rotor wake envelope to delineate the windward undisturbed sampling area, thereby generating a yaw compensation waypoint sequence to control the UAV's climb. The above process utilizes the UAV's native dynamic parameters to lock the native atmospheric physical region free from downwash airflow pollution, and actively tracks this dynamic region through yaw control, thereby eliminating the pollution interference of the aircraft's own flow field on the environmental observation data from the physical space, ensuring the original authenticity of the multi-source atmospheric sampling data.

[0015] (2) A UAV-based vertical profile observation device performs temporal alignment between atmospheric observation sequences and three-dimensional trajectory sequences. It dynamically integrates the sensor's response time constant with the airflow disturbance rate to generate altitude hysteresis compensation parameters, which are then injected back into the vertical axis to reconstruct the hysteresis-corrected altitude sequence. Subsequently, it performs vertical differencing on the matched atmospheric observation sequences, retaining the original sampling density in the extreme abrupt range of the vertical gradient vector and performing feature aggregation and dimensionality reduction in the smooth range. This process reverses the topological restoration of the spatial pose deviation caused by the superposition of hardware sensing delay and aerodynamic disturbances, eliminating altitude distortion in meteorological data during high-speed ascent. Furthermore, it implements non-equidistant sampling reconstruction based on the physical characteristics of atmospheric stratification, achieving effective compression of homogeneous data while accurately characterizing abrupt boundary changes.

[0016] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0017] Figure 1 This is a flowchart of a vertical profile observation method based on an unmanned aerial vehicle (UAV) according to the present invention. Figure 2 This is a schematic diagram illustrating the mapping between the three-dimensional spatial trajectory and observation data of the windward, undisturbed sampling area in an embodiment of the present invention. Figure 3 This is a schematic diagram of an atmospheric vertical profile sequence with non-equidistant sampling in an embodiment of the present invention; Figure 4 This is a flowchart of a vertical profile observation device based on an unmanned aerial vehicle (UAV) according to the present invention. Detailed Implementation

[0018] This application provides a method and apparatus for vertical profile observation based on unmanned aerial vehicles (UAVs), which solves the problems of spatial misalignment and data distortion in atmospheric observation caused by UAV rotor flow field interference and sensor response delay.

[0019] The overall approach of the scheme in this application is as follows: High-frequency inertial vibration sequences and electronically controlled duty cycle sequences of the UAV during its dwell phase are collected. The dominant frequency amplitude vector is extracted and combined with the electronically controlled duty cycle to invert the instantaneous rotor thrust distribution, and the rotor wake envelope is calibrated. The three-axis attitude deflection sequence during the dwell phase is extracted to invert the environmental wind field vector. The environmental wind field vector is then superimposed with the rotor wake envelope using spatial vectors. A windward, undisturbed sampling area is delineated along the outer edge of the wake envelope. The azimuth plane of this sampling area is tracked to generate a yaw compensation waypoint sequence to control the UAV's climb. Simultaneously, atmospheric observation sequences and three-dimensional trajectory sequences are collected. The collected sequences are time-domain aligned. The sensor response time constant is extracted and dynamically integrated with the airflow disturbance rate to generate altitude hysteresis compensation parameters. These parameters are then injected into the vertical axis of the trajectory to reconstruct the hysteresis-corrected altitude sequence. Vertical difference is performed on the atmospheric observation sequences to extract the vertical gradient vector. The original sampling density of meteorological nodes is retained in extreme value abrupt change intervals, and feature aggregation and dimensionality reduction are performed on adjacent nodes in smooth intervals to output a non-equidistant atmospheric vertical profile sequence.

[0020] Example 1; please refer to Figure 1 This invention provides a technical solution: a vertical profile observation method based on an unmanned aerial vehicle (UAV), comprising the following steps: S1, collecting the high-frequency inertial vibration sequence and electronically controlled duty cycle sequence of the UAV during its dwell phase, extracting the dominant frequency amplitude vector and combining it with the electronically controlled duty cycle to invert the instantaneous thrust distribution of the rotor, and calibrating the rotor wake envelope surface under dynamic flow field; S2, extracting the three-axis attitude deflection sequence during the dwell phase to invert the environmental wind field vector, performing spatial vector superposition of the environmental wind field vector and the rotor wake envelope surface, delineating the windward undisturbed sampling area on the outer edge of the rotor wake envelope surface; tracking the azimuth plane of the windward undisturbed sampling area to generate a yaw compensation waypoint sequence, controlling the UAV to climb along the yaw compensation waypoint sequence, and simultaneously collecting atmospheric observations. S3. Perform time-domain alignment on the atmospheric observation sequence and the 3D trajectory sequence, extract the sensor's response time constant, and invert the airflow disturbance rate by combining the rotor's instantaneous thrust distribution in the windward, undisturbed sampling area; dynamically integrate the response time constant and the airflow disturbance rate to generate an altitude hysteresis compensation parameter, and inject it back into the vertical axis of the 3D trajectory sequence to reconstruct the hysteresis-corrected altitude sequence; S4. Perform vertical difference on the atmospheric observation sequence that matches the hysteresis-corrected altitude sequence to extract the vertical gradient vector characterizing the stratification features; retain the original sampling density of meteorological nodes in the extreme value abrupt interval of the vertical gradient vector, and perform feature aggregation and dimensionality reduction on adjacent nodes in the smooth interval to output a non-equidistant atmospheric vertical profile sequence.

[0021] In this implementation scheme, in step S1, when the UAV hovers at the target altitude, the inertial measurement unit inside the flight control system synchronously records the minute vibration data of the aircraft, i.e., the high-frequency inertial vibration sequence of the aircraft. The pulse width modulation signal ratio of the electronic speed controller controlling the motor speed, i.e., the ESC duty cycle sequence, records the real-time output power of the motor. By extracting the main frequency amplitudes from the high-frequency vibration data and combining them with the real-time output power of the motor, the actual thrust generated by each rotor at a specific moment can be calculated. Based on the thrust magnitude, the three-dimensional geometric boundary of the turbulent field formed by the high-speed downward exhaust of the rotor in space can be calculated, i.e., the rotor wake envelope. Defining the specific location and range of the rotor wake envelope under dynamic airflow provides a basic boundary benchmark for subsequent locating air sampling areas unaffected by the UAV exhaust flow.

[0022] In step S2, the angular sequence of the UAV's deviation from the horizontal reference plane in the pitch, roll, and yaw directions during hovering is extracted. Based on the wind-resistant tilting force balance state of the multirotor aircraft, the natural wind direction and speed vector in the current space are calculated backward. The natural wind field vector is synthesized with the three-dimensional geometric boundary of the rotor turbulence field determined in step S1 to confirm the actual affected area after the rotor exhaust is deflected by the external wind. Outside the upwind direction avoiding this affected area, the unmixed native air region, i.e., the undisturbed sampling area, is determined. Based on the spatial orientation of this sampling area as altitude changes, a sequence of flight target points with a specific nose-facing angle is generated. The UAV flies upward along the waypoint sequence with yaw control commands, ensuring that the environmental sensor probes on the fuselage are always pointed at the native air region, and simultaneously recording atmospheric environmental parameters and the corresponding three-dimensional coordinates of the flight space.

[0023] In step S3, the timestamps of the atmospheric data recorded by the environmental sensor and the trajectory coordinate timestamps recorded by the flight control system are first unified to eliminate time reference differences between different hardware devices. The environmental sensor probe has an inherent physical delay time, i.e., a response time constant, from contact with the external gas to the output of a stable electrical signal. Combining this with the local airflow acceleration or deceleration caused by the instantaneous thrust of the rotor at the edge of the sampling area, the actual airflow rate across the sensor probe surface is calculated. The inherent delay time of the sensor is continuously mathematically integrated with the airflow change rate on the probe surface to calculate the spatial misalignment deviation between the recorded environmental data altitude and the actual physical altitude of the air. This spatial misalignment deviation value is then substituted back into the vertical altitude coordinates of the original three-dimensional trajectory for offset correction, calculating the true altitude sequence of the measured air and resolving the altitude distortion of meteorological data during continuous UAV climb observations.

[0024] In step S4, for the atmospheric data sequence after altitude correction, the specific numerical differences in meteorological parameters between adjacent altitude layers are calculated to obtain a set of slopes reflecting the physical structure characteristics of atmospheric vertical stratification, i.e., the vertical gradient vector. For spatial ranges where temperature or pollutant concentration changes drastically with altitude, the high-frequency spatial data nodes originally acquired by the sensor are maintained to ensure the recording accuracy of micro-meteorological features such as inversion layers or pollution boundaries. For spatial ranges of uniformly mixed layers where meteorological parameters change minimally with altitude, multiple adjacent altitude data points with similar physical properties are merged and compressed into a small number of representative nodes. Finally, a vertical sequence map of atmospheric parameters with uneven spatial sampling node spacing is generated, and redundant homogeneous environmental detection data is removed while retaining key details of abrupt changes in key meteorological spatial stratification.

[0025] Specifically, the process of collecting the high-frequency inertial vibration sequence and electronically controlled duty cycle sequence of the UAV during its loitering phase, extracting the dominant frequency amplitude vector, and combining it with the electronically controlled duty cycle to invert the instantaneous rotor thrust distribution, and calibrating the rotor wake envelope surface under the dynamic flow field is as follows: Perform a fast Fourier transform on the high-frequency inertial vibration sequence of the airframe to extract the energy concentration frequency band, and convert the energy concentration frequency band into the rotor rotation speed frequency; fuse the rotor rotation speed frequency and the electronically controlled duty cycle to solve the instantaneous rotor thrust distribution using the lift characteristic matrix; establish the downward exhaust velocity gradient and turbulent dissipation rate distribution boundary based on the instantaneous rotor thrust distribution, map the distribution boundary to the airframe coordinate system, and calibrate the rotor wake envelope surface under the dynamic flow field.

[0026] In this implementation scheme, the high-frequency inertial vibration sequence and electronically controlled duty cycle sequence of the airframe are both legally authorized reads from the underlying hardware of the UAV's local flight control system. The original sensor signals undergo physical desensitization processing at the register level within the microcontroller, and the entire calculation process is performed in network-isolated mode via the local bus. The processing of the high-frequency inertial vibration sequence involves extracting continuously sampled airframe vibration data and discretizing it into the frequency domain to identify the peak coordinates of the power spectral density of the mechanical structure response. The specific calculation logic for converting this concentrated energy frequency band into the rotor speed frequency is as follows: ;in, Indicates the rotor speed frequency; The number of physical blades on one side of the UAV rotor; f represents the search frequency point during the frequency domain transformation process; This indicates the preset lower limit of the rotor operating frequency; This indicates the preset upper limit of the rotor operating frequency; L represents the sampling window length of the high-frequency inertial vibration sequence; n represents the discrete-time index within the sampling window. This represents the vertical vibration acceleration amplitude of the airframe at time index n; i represents the imaginary unit. After obtaining the rotor speed frequency, the system fuses it with the underlying digital control commands of the drive motor using multi-source dynamic states. Since the rotor thrust depends not only on its steady-state rotational angular velocity but also on the feedforward forcing effect of the instantaneous large current injection from the electronic speed controller, the calculation logic for fusing these two factors into the lift characteristic matrix to solve for the instantaneous rotor thrust distribution is as follows: ;in, This represents the instantaneous thrust distribution value of the p-th rotor; p represents the rotor physical number; Q represents the total number of airborne power motors; q represents the stator coil index variable inside the motor. This represents the aerodynamic lift coefficient in the lift characteristic matrix that reflects the rotational speed response; This represents the empirical constant of the thrust exponent based on the physical shape of the duct. This represents the torque compensation coefficient in the lift characteristic matrix that reflects the ESC control response; This represents the current ESC duty cycle of the q-th motor. The key parameter is the aerodynamic lift coefficient. The determination method is as follows: The static thrust calibration value of the dynamic thrust gauge is read on a multi-rotor fixed-axis dynamic test bench. The least squares method is used to identify the physical parameters of multiple sets of rotational speed characteristic data matrices. The aerodynamic slope constant under the corresponding blade geometric twist angle structure is extracted and directly assigned as the coefficient. Since the velocity gradient and turbulent dissipation effect formed by downward exhaust exhibit exponential nonlinear decay with increasing spatial depth, the instantaneous thrust distribution is mapped to the airframe coordinate system as a source term. The boundary discrimination logic for the rotor wake envelope surface under the calibrated dynamic flow field is as follows: ;in, Represents a point in three-dimensional space in the body coordinate system. The boundary discrimination value of the turbulent dissipation rate distribution at a certain point is determined if and only if the calculated value is greater than zero, indicating that the spatial coordinate is inside the rotor wake envelope. Represents the lateral distance measurement coordinates of the body coordinate system; Represents the longitudinal distance measurement coordinates of the body coordinate system; Represents the vertical depth coordinates of the body coordinate system; This represents the physical correction factor for spatial coupling interference in the flow field of a multi-rotor UAV. Its physical meaning lies in the fact that the downwash airflow from adjacent rotors of a multi-rotor UAV collides and interferes in space, leading to a surge in local flow velocity. The determination method involves calculating the ratio of the center distance between adjacent rotors to the blade radius, and then calibrating the value based on the peak wind speed amplification factor in the interference region measured by wind tunnel experiments. This indicates the real-time air density obtained from the airborne weather probe; This represents the equivalent swept area of ​​a single rotor on the horizontal plane. The variance parameter representing the outward diffusion distribution of the downwash flow, whose physical meaning is to quantify the cone angle size of the downwash flow spreading around with the increase of depth. Determination method: By using smoke flow field visualization to photograph the boundary of the downwash flow in the static hovering state, and calculating the statistical variance by extracting the expansion slope of the cross-sectional area of the airflow boundary changing with depth; Represents the critical threshold of the turbulent dissipation rate for dividing the wake envelope surface. Regarding the critical threshold of the turbulent dissipation rate The determination method is as follows: In the flow field calibration experiment of the standard environmental wind tunnel, use a hot-wire anemometer array to measure the background turbulent kinetic energy of the original atmosphere under different calm wind conditions, extract the peak value of the upper envelope line of the background turbulent kinetic energy distribution of the undisturbed atmosphere, and after deducting the sensor thermal noise baseline, use it as the dynamic value of this threshold.

[0027] Specifically, the process of extracting the triaxial attitude angle sequence in the residence stage to invert the environmental wind field vector, performing spatial vector superposition of the environmental wind field vector and the rotor wake envelope surface, and demarcating the upwind undisturbed sampling area along the outer edge of the rotor wake envelope surface is as follows: Decompose the triaxial attitude angle sequence to extract the pitch component and the roll component, and convert the pitch component and the roll component into a horizontal drag compensation vector according to the hovering force balance equation; fuse the horizontal drag compensation vector and the instantaneous thrust distribution of the rotor to invert the environmental wind field vector; extract the offset projection boundary of the environmental wind field vector on the rotor wake envelope surface, and intercept the characteristic space that is not sheared by the downwash flow field along the upwind outer edge of the offset projection boundary to demarcate the upwind undisturbed sampling area.

[0028] In this implementation plan, when the system extracts the triaxial attitude angle sequence in the residence stage, it directly reads the original fused data of the gyroscope and accelerometer through the serial peripheral interface bus at the bottom layer of the flight control main board, and completes the physical signal desensitization at the microcontroller register level to ensure the legality and isolation security of the data source. After decomposing the attitude angle sequence to obtain the pitch component and the roll component, substitute them into the hovering force balance equation to convert them into a horizontal drag compensation vector. Specifically, when the UAV maintains a fixed point in the wind, the partial aerodynamic pull generated by the tilting of the fuselage is exactly equal to the environmental wind resistance. The calculation logic for inverting the environmental wind field vector by combining the overall aerodynamic force calibrated in the early stage is as follows: ; where Represents the environmental wind field vector in the two-dimensional plane; Represents the total aerodynamic pull of the multi-rotor fuselage vertically upward; Represents the pitch angle component at the discrete sampling node; Represents the roll angle component at the same sampling node; Represents the measured density of the external ambient atmosphere; Represents the aerodynamic drag coefficient of the UAV fuselage in the current upwind configuration; Represents the equivalent projected cross-sectional area of the UAV fuselage along the wind direction; This represents an infinitesimal compensation constant set to prevent division-to-zero overflow during hovering in calm conditions. It is relevant to the key parameter, the aerodynamic drag coefficient. The determination method is as follows: Place the entire drone of the same model in an environmental calibration wind tunnel, record the fuselage tilt angle and the drag data output by the bottom force balance under the preset wind speed gradient level, and in the environmental calibration wind tunnel, traverse the drone at pitch angles with a fixed step size (e.g., 5°). Roll angle All attitude combinations within the range are recorded, along with the three-dimensional drag component data output by the underlying force balance. Subsequently, a bivariate Gaussian surface fitting algorithm is used to smooth the discrete drag data, generating a continuous aerodynamic drag coefficient surface. This surface is then discretized and written into the flash memory of the flight control chip, constructing a static lookup table mapping three-dimensional tilt angles to drag coefficients. During computation, lookup interpolation calls are executed. After acquiring the environmental wind field vector, the system performs spatial vector superposition to delineate the undisturbed upwind sampling area. The system performs dynamic offset projection calculations on the rotor wake envelope surface under calm wind conditions using the environmental wind field vector. The logic for defining the feature space unaffected by downwash flow shearing at the outward edge of the offset projection boundary in the upwind direction is as follows: ;in, This represents the set of three-dimensional coordinates of the defined, undisturbed windward sampling area. Indicates the spatial candidate coordinate points; This represents the wind speed entrainment mapping constant that indicates the wind field disturbance as it evolves with vertical spatial depth. and These represent the environmental wind field vectors respectively. Scalar components in the transverse and longitudinal directions of the body; This indicates the envelope of the unbiased front rotor wake at the corresponding depth. The physical radius at that location; This represents the fluid isolation safety margin set to prevent boundary layer turbulent shear extension; where the wake centerline will tilt under the influence of the wind field, and the offset amount... With vertical depth They are positively correlated. Simultaneously, the wake radius... It is also a function of depth. By introducing a depth variable, the original two-dimensional disk is expanded into a three-dimensional diffusion frustum tilted downwind, thereby accurately defining the undisturbed region in three-dimensional space.

[0029] Specifically, the process of generating a yaw-compensated waypoint sequence by tracking the azimuth plane of the undisturbed windward sampling area, controlling the UAV to climb along the yaw-compensated waypoint sequence, and simultaneously acquiring atmospheric observation sequences and three-dimensional trajectory sequences is as follows: Calculate the dynamic center coordinates of the undisturbed windward sampling area on the vertical trajectory; establish yaw constraint parameters by associating the dynamic center coordinates with the angle between the incoming environmental wind field vector and the wind field vector, generating a yaw-compensated waypoint sequence containing four dimensions: longitude, latitude, altitude, and yaw angle; drive the UAV to execute the yaw-compensated waypoint sequence, triggering the hardware-level clock synchronization latch signal between the atmospheric observation sensor acquisition end and the positioning module output end during the climb process, and acquiring atmospheric observation sequences and three-dimensional trajectory sequences.

[0030] In this implementation scheme, when generating a yaw compensation waypoint sequence by tracking the azimuth plane of the undisturbed windward sampling area, the system first extracts the geometric centroid of the sampling area on each vertical climb section of the three-dimensional flight trajectory, and calculates the dynamic center coordinates of the undisturbed windward sampling area on the vertical trajectory. Subsequently, the system correlates the dynamic center coordinates with the angle between the incoming environmental wind field vector and establishes the calculation logic for the yaw constraint parameters that can force the sensor probe to be in the absolute windward position: ;in, Indicates the first Yaw angle control commands at each spatial climb step node; Discrete node indices representing the climbing trajectory; and This represents the dynamic center coordinates of the undisturbed sampling area corresponding to the node. and Represents the x and y coordinates of the fuselage's geometric center at the current trajectory node; The smooth tracking coefficient represents the yaw response of the UAV in a dynamic turbulent wind field; This represents the environmental wind field vector refreshed at the current height node; This represents the displacement vector of the atmospheric observation sensor probe relative to its fixed installation directly in front of the aircraft nose. (This is related to the smoothing tracking coefficient.) The determination method is as follows: read the yaw moment of inertia and peak torque of the motor at the bottom layer of the flight control board, and select the reciprocal of the response time required for the yaw drive system to reach the rated angular velocity as the value to prevent high-frequency mechanical oscillations during heading tracking. After outputting a yaw compensation waypoint sequence containing four dimensions of longitude, latitude, altitude, and yaw angle, the UAV is driven to climb along the yaw compensation waypoint sequence. During the climb process, to avoid communication delays caused by operating system process scheduling, the precision timer at the bottom layer of the flight control is directly configured to external hard interrupt mode. The rising edge of the global positioning second pulse output by the positioning module is used as a direct trigger source to send a hardware-level clock synchronization latch signal to the analog-to-digital converter pin of the atmospheric observation sensor along the hardware circuit. This hardware deep collaboration mechanism forces multi-source observation devices to solidify their level states within the same microsecond-level time slice, synchronously acquiring atmospheric observation sequences and three-dimensional trajectory sequences, eliminating the spatiotemporal misalignment error of multi-sensor data fusion from the physical bottom layer, and achieving absolute rigid alignment of environmental parameters and trajectory pose.

[0031] Specifically, the process of performing time-domain alignment on the atmospheric observation sequence and the three-dimensional trajectory sequence, extracting the sensor's response time constant, and inverting the airflow disturbance rate by combining the rotor's instantaneous thrust distribution in the undisturbed upwind sampling area is as follows: Extract the hardware clock markers of the atmospheric observation sequence and the three-dimensional trajectory sequence; evaluate the clock drift offset between the hardware clock markers using a cross-correlation function; perform phase-compensated resampling to output time-domain aligned observation trajectory synchronization data; analyze the probe thermal constant and gas diffusion flux of the environmental sensor and combine them with the current initial environmental state to extract the response time constant of the nonlinear response interval; map the rotor's instantaneous thrust distribution to the boundary of the undisturbed upwind sampling area, calculate the flow field kinetic energy attenuation coefficient at the boundary, and solve the forced convection acceleration component of the ambient atmosphere based on the flow field kinetic energy attenuation coefficient to invert and generate the airflow disturbance rate.

[0032] In this implementation scheme, the underlying hardware clock markers for the atmospheric observation sequence and the 3D trajectory sequence are directly read from the sensor pins via the asynchronous transceiver bus of the UAV motherboard. This process directly completes the timestamp mapping and data anonymization in the microcontroller's underlying registers, without accessing external wide area networks or calling cloud APIs. This establishes the legitimate source of the original detection data at the physical level and provides a system fallback mechanism to prevent data leakage. Time-domain alignment uses a cross-correlation function to calculate the phase offset of the pulse waveform, reconstructing the aligned trajectory data based on the high-precision second pulse of the positioning module. The calculation process for extracting the response time constant based on the physical delay analysis of the environmental sensors and the current initial environmental state is as follows: ;in, The response time constant represents the nonlinear response range; This indicates the physical quality of the sensor's core probe; Indicates the nominal specific heat capacity of the probe's sensing element material; This represents the effective sensing surface area of ​​the probe in direct contact with the airflow. This represents the basic natural diffusion flux of ambient air entering the sensor's measurement cavity; This represents the coupling perturbation coefficient of the current environmental temperature and humidity physical state on the gas viscous drag. The system then uses the previously inverted instantaneous thrust distribution as a source term, mapping it to the physical boundary of the unperturbed upwind sampling area to evaluate the fluid's kinetic energy decay characteristics at this location. The logic for solving the forced convection acceleration component of the ambient atmosphere and inverting the airflow perturbation rate is as follows: ;in, This represents the airflow disturbance rate at the synchronization time node t_s; This represents the kinetic energy attenuation coefficient of the flow field at the boundary; This represents the total instantaneous thrust scalar mapped to the windward sampling boundary surface; This indicates the local air density read in real time by the sensor; This represents the forced convection acceleration component driven by the pressure difference in the rotor exhaust. According to Bernoulli's equation and the fluid momentum theorem, the pressure difference in the exhaust from the upper and lower surfaces of the rotor drives the static air to accelerate downwards. By extracting local dynamic pressure data measured by pressure sensors and pitot tubes, and using the assumption of incompressible fluid, the static pressure drop at the rotor plane is converted into the forced convection acceleration component at the flow field boundary. Flow field kinetic energy attenuation coefficient The method for determining it is as follows: In the standard wind tunnel calibration experiment before flight, the fluid attenuation profile of the rotor under different duty cycles is measured using a hot-wire anemometer array, and the distance attenuation base at which the wind speed kinetic energy at the boundary layer drops below the natural background wind speed limit is extracted as the constant.

[0033] Specifically, the process of generating height hysteresis compensation parameters by dynamically integrating the response time constant and airflow disturbance rate, and then back-injecting them into the vertical axis of the three-dimensional trajectory sequence to reconstruct the hysteresis-corrected height sequence is as follows: A differential equation containing the convection terms of the response time constant and airflow disturbance rate is constructed. Numerical continuous integration is performed on the differential equation along the time axis, and the height hysteresis compensation parameters are discretized and output. The original height coordinate nodes on the vertical axis of the three-dimensional trajectory sequence are extracted. The height hysteresis compensation parameters are converted into negative offset coordinate vectors. A spatial inverse topological substitution is performed between the negative offset coordinate vectors and the original height coordinate nodes. The topological substitution result is mapped back to the original height coordinates, and the hysteresis-corrected height sequence is reconstructed.

[0034] In this implementation scheme, the sensor's physical delay and rotor airflow disturbance jointly determine the spatial misalignment of the observed data. The sensor's perception of gas parameters is a thermodynamic / kinetic relaxation process, and its measured values ​​lag behind changes in the actual environment. When constructing the differential equation, the sensor's inherent response time constant is considered as the system's inertial term, and the airflow disturbance rate is considered as a nonlinear acceleration term of the gas exchange process due to external forced convection. Continuous integration along the time axis essentially calculates the additional vertical spatial distance the UAV flies due to the measurement delay during its climb. The process of constructing a differential equation containing the response time constant and the airflow disturbance rate convection term, and then performing discretized numerical continuous integration along the time axis to generate the altitude hysteresis compensation parameters is as follows: ;in, represents the height hysteresis compensation parameter output at the c-th trajectory node; c represents the discrete spatial node index after completing time-domain alignment; w represents the sliding summation variable in the numerical integration process; This represents the physical vertical climb velocity of the drone at the corresponding spatial node; This represents the environmental response time constant dynamically mapped to this node; This represents the airflow disturbance rate at the corresponding node boundary; This represents the reference forced convection constant of the sensor's measurement cavity; This represents the fixed resampling time step after hardware clock alignment. Key parameter: Reference forced convection constant. The determination method is as follows: Place the corresponding sensor module in a laminar flow calibration chamber, gradually increase the inlet flow velocity, and when the sensor step response time converges and stabilizes at the minimum limit threshold specified in the hardware datasheet, record the forced airflow velocity in the calibration chamber as this value. After obtaining the altitude hysteresis compensation parameters, the system extracts the original vertical trajectory data from the GNSS and barometer fusion calculations, converts it into a reverse offset state, and performs spatial inverse topology substitution. The mapping logic for reconstructing the hysteresis correction altitude sequence is as follows: ;in, Represents the original height coordinates remapped after topology substitution; This represents the vertical height measurement reading of the original three-dimensional trajectory sequence at this node; This represents the instantaneous physical deflection angle of the UAV's onboard sensor mount relative to the absolute vertical line of gravity. This process forcibly reverts hysteresis-lagging meteorological data to its original physical altitude in the underlying three-dimensional coordinate system, fundamentally correcting the nonlinear slippage of the vertical profile altitude sequence under continuous sensing conditions. (See also...) Figure 2As shown in the figure, the three-dimensional coordinate system consists of "horizontal," "vertical," and "height," constructing the physical detection space of the UAV. The spiraling solid line in the figure represents the yaw-compensated waypoint sequence climb trajectory dynamically generated by the UAV in a complex wind field environment to continuously and actively track the undisturbed sampling area facing the wind. The spatial scattered points distributed on the trajectory represent multi-source observation nodes acquired simultaneously. It is worth noting that the coordinates of these scattered points on the vertical height axis are not the original air pressure or GPS altimeter, but the true native height reconstructed after back-injection with altitude hysteresis compensation parameters applied, intuitively reflecting the system's correction effect on spatial misalignment caused by sensor response delay. In addition, the color gradient feature of the scattered points maps the intensity distribution of the characteristic values ​​of specific meteorological observation elements, fully demonstrating the three-dimensional meteorological data matching status under the condition of no downwash airflow disturbance.

[0035] Specifically, the process of performing vertical differencing on the atmospheric observation sequence matching the hysteresis-corrected altitude sequence to extract the vertical gradient vector characterizing the stratification features is as follows: The hysteresis-corrected altitude sequence and the atmospheric observation sequence are subjected to spatial coordinate mapping to construct a spatial altitude-meteorological element correlation matrix; an adaptive variable step-size sliding window is configured along the spatial altitude dimension of the spatial altitude-meteorological element correlation matrix; within the adaptive variable step-size sliding window, higher-order forward differencing is performed on the meteorological elements to extract the slope fluctuation envelope of the meteorological elements as a function of altitude; the slope fluctuation envelope is then converted into a vertical gradient vector containing the stratification temperature rate sequence and the substrate concentration jump rate sequence.

[0036] In this implementation scheme, the system performs spatial coordinate mapping in the airborne edge computing unit and directly completes meteorological data processing in the local cache, ensuring the real-time performance and computational efficiency of environmental detection data. When performing spatial coordinate mapping between the control hysteresis-corrected altitude sequence and the atmospheric observation sequence, the system binds the hardware-aligned altitude cursor to multi-dimensional meteorological data using key-value pairs and constructs a spatial altitude-meteorological element correlation matrix in memory. The logic for configuring an adaptive variable-step-size sliding window along the spatial altitude dimension and performing high-order forward differencing on meteorological elements within this window is as follows: ;in, This represents the slope fluctuation envelope value extracted at the corrected height node g; g represents the discrete index of the height sequence after spatial mapping. This represents the physical span of the sliding window dynamically allocated at this node based on the local data acquisition density; Indicates the forward difference order of the configuration; r represents the offset step offset during the higher-order difference expansion process; This represents the actual meteorological element observation values ​​stored at the offset node g+r. After acquiring the slope fluctuation envelope, the system physically separates it according to the data channels of the internal hardware sensors, converting the difference sequences of the temperature sensing channel and the specific pollutant gas sensing channel into multidimensional vertical gradient vectors containing the stratification temperature change rate sequence and the substrate concentration jump rate sequence, thereby quantifying the intensity of atmospheric physical abrupt changes in different altitude ranges at the digital level.

[0037] Specifically, the process of preserving the original sampling density of meteorological nodes in the extreme abrupt change interval of the vertical gradient vector, and performing feature aggregation and dimensionality reduction on adjacent nodes in the smooth interval to output a non-equidistant atmospheric vertical profile sequence is as follows: Calculate the first derivative of the vertical gradient vector, locate the abrupt boundary layer region where the absolute value of the derivative exceeds the limit and the uniform mixing layer region where the derivative approaches zero, label the abrupt boundary layer region as the extreme abrupt change interval, and label the uniform mixing layer region as the smooth interval; lock the meteorological nodes within the extreme abrupt change interval and preserve the original sampling density; extract the sequences of adjacent meteorological nodes within the smooth interval, configure an adaptive spatial distance clustering algorithm to calculate the physical feature centroid of adjacent node clusters, and extract the feature centroid as the representative node for dimensionality reduction; splice the meteorological nodes within the extreme abrupt change interval with the representative node for dimensionality reduction, and reorder them according to the spatial height index to generate a non-equidistant atmospheric vertical profile sequence.

[0038] In this implementation scheme, when calculating the first derivative of the vertical gradient vector to locate the abrupt boundary layer region and the uniform mixing layer region, the system performs an absolute value limit comparison in the underlying microprocessor logic gate array. After locking the meteorological nodes within the extreme value abrupt interval, all the original sampling time series and spatial topological states corresponding to this physical interval are fully sealed in non-volatile memory, preserving the original sampling density to accurately depict the microscopic contours of the inversion layer. The extraction logic for extracting adjacent meteorological node sequences within the smooth interval and configuring an adaptive spatial distance clustering algorithm to calculate the physical feature centroid is as follows: ;in, This represents the multidimensional reduced-dimensional representative node vector extracted by distance weighting within a smooth interval; represents the set of meteorological node indices that are classified into a uniformly mixed layer region based on derivatives; b represents the local search index variable within the set; This indicates the original multidimensional meteorological feature node invoked at search index b; This represents the center of the arithmetic and geometric mean of all meteorological feature nodes within the flat interval; This parameter represents the standard deviation of the physical distribution of meteorological characteristic data within the region. The method for determining the threshold for exceeding the absolute value of the derivative is as follows: During the near-Earth boundary layer preheating phase before UAV takeoff, the system collects a white noise baseline sequence of the background atmosphere, calculates the root mean square error of the first derivative of the baseline, and extracts three times this root mean square error as the adaptive threshold for exceeding the limit. After reading the dimension-reduced representative nodes within the smooth intervals, the system performs bottom-level data block concatenation with the full set of meteorological nodes retained in the extreme value abrupt change intervals. Based on the spatial height index coordinates attached to the meteorological nodes, it re-executes ascending order rearrangement to generate a non-equidistant atmospheric vertical profile sequence that balances the fidelity of microscopic abrupt boundary changes with the lightweighting of global data. (See also...) Figure 3 ,against Figure 3 The labels appearing in the figure are specifically defined here: "Corrected Height" refers to the original height coordinate system reconstructed after temporal alignment and hysteresis compensation parameter back-injection; "Ambient Temperature" (corresponding to the bottom horizontal axis label "Temperature") and "Aerosol Concentration" (corresponding to the top horizontal axis label "Concentration") represent two typical atmospheric feature elements extracted from the bottom layer of the system, respectively. As shown in the figure, the solid line (with circular nodes) and the dashed line (with square nodes) depict the vertical difference variation trend of these two elements with the corrected height, respectively. In the spatial interval where meteorological features undergo drastic changes (i.e., the extreme value mutation interval where the horizontal slope of the curve fluctuates greatly), the curve retains high-frequency and dense data sampling nodes, thereby faithfully restoring the microscopic mutation features such as the inversion layer and the pollution boundary layer; while in the relatively stable uniform mixing layer of meteorological features (i.e., the gentle interval where the curve tends to be steep), the system performs feature aggregation and dimensionality reduction through clustering algorithms, making the distribution of representative nodes in this interval extremely sparse. This figure clearly verifies the final effect of this scheme in balancing the preservation of stratification mutation features and global homogeneous data compression.

[0039] Example 2; please refer to Figure 4A vertical profile observation device based on a UAV is used to execute a vertical profile observation method based on a UAV as described in the embodiments. The device includes: a wake calibration unit, used to collect the high-frequency inertial vibration sequence and electronically controlled duty cycle sequence of the UAV during its dwell phase, extract the dominant frequency amplitude vector, and combine it with the electronically controlled duty cycle to invert the instantaneous thrust distribution of the rotor, calibrating the rotor wake envelope surface under dynamic flow field; a trajectory acquisition unit, used to extract the three-axis attitude deflection sequence during the dwell phase to invert the environmental wind field vector, superimpose the environmental wind field vector with the rotor wake envelope surface using spatial vectors, and delineate a windward undisturbed sampling area on the outer edge of the rotor wake envelope surface; and to generate a yaw compensation waypoint sequence by tracking the azimuth plane of the windward undisturbed sampling area, controlling the UAV to climb along the yaw compensation waypoint sequence, and simultaneously acquiring... The system integrates atmospheric observation sequences and 3D trajectory sequences. A hysteresis correction unit performs temporal alignment between the atmospheric observation sequences and the 3D trajectory sequences, extracts the sensor's response time constant, and inverts the airflow disturbance rate by combining the rotor's instantaneous thrust distribution in the windward, undisturbed sampling area. It dynamically integrates the response time constant and the airflow disturbance rate to generate a height hysteresis compensation parameter, which is then injected back into the vertical axis of the 3D trajectory sequence to reconstruct the hysteresis-corrected height sequence. A profile generation unit performs vertical differencing on the atmospheric observation sequences that match the hysteresis-corrected height sequence, extracting vertical gradient vectors that characterize stratification features. It retains the original sampling density of meteorological nodes in the extreme value abrupt range of the vertical gradient vector, and performs feature aggregation and dimensionality reduction on adjacent nodes in the smooth range, outputting a non-equidistant atmospheric vertical profile sequence.

[0040] In this implementation scheme, the wake calibration unit reads the high-frequency inertial vibration sequence and electronically controlled duty cycle sequence of the UAV during its stay at a specified altitude in real time via the onboard bus. This unit performs a frequency domain transformation on the high-frequency inertial vibration sequence to extract the dominant frequency amplitude vector, and then fuses the dominant frequency amplitude vector with the underlying electronically controlled duty cycle to calculate the real-time instantaneous thrust distribution of each rotor. Subsequently, based on the rotor's instantaneous thrust distribution, the unit establishes the turbulent dissipation boundary for downward exhaust, maps the spatial coordinates of this boundary to the UAV's body coordinate system, and outputs the range of the rotor wake envelope under the dynamic flow field.

[0041] The trajectory acquisition unit calls upon the three-axis attitude deflection sequence at the bottom layer of the flight control system and uses the hovering force balance equation to inversely deduce the environmental wind field vector in the current space. This unit performs a spatial offset projection superposition operation on the environmental wind field vector and the rotor wake envelope surface, and extracts the physical space unaffected by the downwash airflow along the windward outer edge of the offset projection boundary, defining it as the windward undisturbed sampling area. Next, this unit calculates the dynamic center of the windward undisturbed sampling area and generates a yaw compensation waypoint sequence by combining it with the incoming environmental wind field angle. It then issues trajectory execution commands to the flight control module to control the UAV to climb, and triggers a clock synchronization latch signal during the climb, simultaneously recording the atmospheric observation sequence and the three-dimensional trajectory sequence.

[0042] The hysteresis correction unit extracts the hardware clock markers from the atmospheric observation sequence and the 3D trajectory sequence, and performs phase-compensated resampling to achieve absolute alignment of the two sequences in the time domain. This unit analyzes the physical parameters of the environmental sensors to extract the response time constant and maps the instantaneous rotor thrust distribution to calculate the airflow disturbance rate at the boundary of the unperturbed sampling area. This unit constructs a differential equation containing the response time constant and the airflow disturbance rate and continuously integrates along the time axis, discretizing the output altitude hysteresis compensation parameter. This unit converts the altitude hysteresis compensation parameter into a negative offset coordinate vector and performs a spatial inverse topological substitution with the original altitude coordinate nodes on the vertical axis of the 3D trajectory sequence to reconstruct the hysteresis-corrected altitude sequence.

[0043] The profile generation unit controls the hysteresis-corrected altitude sequence and atmospheric observation sequence to perform spatial coordinate mapping to establish an association matrix. Within the association matrix, an adaptive variable-step-size sliding window is configured to perform high-order forward differencing on meteorological elements, extracting vertical gradient vectors characterizing atmospheric stratification features. This unit calculates the first derivative of the vertical gradient vector to distinguish the physical state of atmospheric stratification. Within the range determined to be an extreme value abrupt change, the original sampling density of meteorological nodes is retained as is. Within the range determined to be a smooth range, this unit calls a clustering algorithm to calculate the centroids of the physical features of adjacent meteorological node clusters as representative nodes for dimensionality reduction. Finally, the meteorological nodes and dimensionality-reduced representative nodes within the extreme value abrupt change range are rearranged according to spatial altitude index to generate a non-equidistant atmospheric vertical profile sequence.

[0044] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0045] This invention is described with reference to flowchart illustrations and / or block diagrams of systems, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0046] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0047] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0048] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0049] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for observing vertical profiles based on unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: S1. Collect the high-frequency inertial vibration sequence and electronically controlled duty cycle sequence of the UAV during the dwell phase, extract the main frequency amplitude vector and combine it with the electronically controlled duty cycle to invert the instantaneous thrust distribution of the rotor, and calibrate the rotor wake envelope surface under the dynamic flow field. S2. Extract the three-axis attitude deflection sequence during the dwell phase to invert the environmental wind field vector, and perform spatial vector superposition between the environmental wind field vector and the rotor wake envelope. Delineate the windward undisturbed sampling area on the outer edge of the rotor wake envelope. Track the azimuth plane of the windward undisturbed sampling area to generate a yaw compensation waypoint sequence, control the UAV to climb along the yaw compensation waypoint sequence, and simultaneously collect atmospheric observation sequences and three-dimensional trajectory sequences. S3. Perform time-domain alignment on the atmospheric observation sequence and the three-dimensional trajectory sequence, extract the sensor's response time constant, and invert the airflow disturbance rate by combining the rotor's instantaneous thrust distribution in the windward undisturbed sampling area; dynamically integrate the response time constant and the airflow disturbance rate to generate an altitude hysteresis compensation parameter, and inject it into the vertical axis of the three-dimensional trajectory sequence to reconstruct the hysteresis-corrected altitude sequence. S4. Perform vertical difference on the atmospheric observation sequence of the matched hysteresis-corrected height sequence to extract the vertical gradient vector characterizing the stratification features; retain the original sampling density of meteorological nodes in the extreme value abrupt interval of the vertical gradient vector, and perform feature aggregation and dimensionality reduction on adjacent nodes in the smooth interval to output the atmospheric vertical profile sequence with non-equidistant sampling.

2. The method for vertical profile observation based on unmanned aerial vehicles according to claim 1, characterized in that: The specific process of collecting high-frequency inertial vibration sequences and electronically controlled duty cycle sequences of the UAV during its loitering phase, extracting the dominant frequency amplitude vector, and combining it with the electronically controlled duty cycle to invert the instantaneous rotor thrust distribution, and calibrating the rotor wake envelope surface under dynamic flow field is as follows: The high-frequency inertial vibration sequence of the airframe is subjected to fast Fourier transform to extract the concentrated energy frequency band, which is then converted into the rotor speed frequency. The instantaneous thrust distribution of the rotor is calculated by integrating the rotor speed frequency and the electronically controlled duty cycle to the lift characteristic matrix. Based on the instantaneous thrust distribution of the rotor, establish the velocity gradient of the downward exhaust and the boundary of the turbulent dissipation rate distribution, map the distribution boundary to the airframe coordinate system, and calibrate the rotor wake envelope surface under the dynamic flow field.

3. The method for vertical profile observation based on unmanned aerial vehicles according to claim 1, characterized in that: The process of extracting the three-axis attitude deflection sequence during the dwell phase to invert the environmental wind field vector, superimposing the environmental wind field vector with the rotor wake envelope surface using spatial vector superposition, and delineating the windward undisturbed sampling area on the outer edge of the rotor wake envelope surface is as follows: The pitch and roll components are extracted by decomposing the three-axis attitude deflection sequence, and then converted into horizontal drag compensation vectors according to the hovering force balance equation. The environmental wind field vector is obtained by fusing the horizontal drag compensation vector with the instantaneous thrust distribution of the rotor; The offset projection boundary of the environmental wind field vector onto the rotor wake envelope is extracted. The feature space unaffected by downwash flow shear is intercepted at the outer edge of the offset projection boundary in the windward direction, and the undisturbed sampling area in the windward direction is delineated.

4. A method for vertical profile observation based on an unmanned aerial vehicle (UAV) according to claim 1 or 3, characterized in that: The specific process of generating a yaw-compensated waypoint sequence by tracking the azimuth plane of the windward, undisturbed sampling area, controlling the UAV to climb along the yaw-compensated waypoint sequence, and simultaneously collecting atmospheric observation sequences and three-dimensional trajectory sequences is as follows: Calculate the dynamic center coordinates of the windward, undisturbed sampling area on the vertical trajectory; By associating the dynamic center coordinates with the angle between the incoming wind field vector and the environmental wind field, yaw constraint parameters are established, generating a yaw compensation waypoint sequence containing four dimensions: longitude, latitude, altitude, and yaw angle. The drone is driven to execute a yaw compensation waypoint sequence. During the climb process, the hardware-level clock synchronization latch signal between the atmospheric observation sensor acquisition end and the positioning module output end is triggered to acquire atmospheric observation sequences and three-dimensional trajectory sequences.

5. The method for vertical profile observation based on unmanned aerial vehicles according to claim 1, characterized in that: The specific process of performing time-domain alignment between the atmospheric observation sequence and the three-dimensional trajectory sequence, extracting the sensor's response time constant, and inverting the airflow disturbance rate by combining the instantaneous rotor thrust distribution in the undisturbed upwind sampling area is as follows: Extract the hardware clock markers from the atmospheric observation sequence and the three-dimensional trajectory sequence, evaluate the clock drift bias between the hardware clock markers through the cross-correlation function, perform phase compensation resampling, and output time-domain aligned observation trajectory synchronization data. By analyzing the probe thermal constant and gas diffusion flux of the environmental sensor and combining them with the current initial environmental state, the response time constant of the nonlinear response range is extracted. The instantaneous thrust distribution of the rotor is mapped to the boundary of the windward undisturbed sampling area. The flow field kinetic energy attenuation coefficient at the boundary is calculated. Based on the flow field kinetic energy attenuation coefficient, the forced convection acceleration component of the ambient atmosphere is solved, and the airflow disturbance rate is generated by inversion.

6. A vertical profile observation method based on an unmanned aerial vehicle (UAV) according to claim 1 or 5, characterized in that: The specific process of generating a height hysteresis compensation parameter by dynamically integrating the response time constant with the airflow disturbance rate, and then back-injecting it into the vertical axis of the three-dimensional trajectory sequence to reconstruct the hysteresis-corrected height sequence is as follows: A differential equation containing the response time constant and the convection term of the airflow disturbance rate is constructed. The differential equation is then continuously integrated numerically along the time axis, and the height hysteresis compensation parameter is discretized and output. The original height coordinate nodes on the vertical axis of the 3D trajectory sequence are extracted. The height hysteresis compensation parameter is converted into a negative offset coordinate vector. The negative offset coordinate vector is driven to perform a spatial inverse topological substitution with the original height coordinate nodes. The topological substitution result is mapped to the original height coordinates to reconstruct the hysteresis-corrected height sequence.

7. The method for vertical profile observation based on unmanned aerial vehicles according to claim 1, characterized in that: The specific process of performing vertical differencing on the atmospheric observation sequence of the matched hysteresis-corrected height sequence to extract the vertical gradient vector characterizing the stratification features is as follows: The spatial coordinate mapping between the controlled hysteresis-corrected altitude sequence and the atmospheric observation sequence is performed to construct a spatial altitude-meteorological element correlation matrix; An adaptive variable step size sliding window is configured on the spatial height-meteorological element correlation matrix along the spatial height dimension. Within the adaptive variable step size sliding window, a high-order forward difference is performed on the meteorological elements to extract the slope fluctuation envelope of the meteorological elements as they change with height. The slope fluctuation envelope is converted into a vertical gradient vector containing the sequence of stratification temperature change rate and the sequence of substrate concentration jump rate.

8. A method for vertical profile observation based on an unmanned aerial vehicle (UAV) according to claim 1 or 7, characterized in that: The specific process of preserving the original sampling density of meteorological nodes in the extreme value abrupt interval of the vertical gradient vector, and performing feature aggregation and dimensionality reduction on adjacent nodes in the smooth interval to output the non-equidistant atmospheric vertical profile sequence is as follows: Calculate the first derivative of the vertical gradient vector, locate the abrupt boundary layer region where the absolute value of the derivative exceeds the limit and the uniform mixing layer region where the derivative approaches zero, and label the abrupt boundary layer region as the extreme value abrupt interval and the uniform mixing layer region as the smooth interval. Meteorological nodes within the extreme value mutation range retain their original sampling density; Extract the sequences of adjacent meteorological nodes within the smooth interval, configure an adaptive spatial distance clustering algorithm to calculate the physical feature centroids of adjacent node clusters, and extract the feature centroids as representative nodes for dimensionality reduction; Meteorological nodes within extreme value abrupt change intervals are spliced ​​together with dimension-reduced representative nodes, and then reordered according to spatial height index to generate a non-equidistant atmospheric vertical profile sequence.

9. A UAV-based vertical profile observation device, used to execute the UAV-based vertical profile observation method according to any one of claims 1-8, characterized in that, include: The wake calibration unit is used to collect the high-frequency inertial vibration sequence and electronically controlled duty cycle sequence of the UAV during the hovering phase, extract the main frequency amplitude vector and combine it with the electronically controlled duty cycle to invert the instantaneous thrust distribution of the rotor, and calibrate the rotor wake envelope surface under the dynamic flow field. The trajectory acquisition unit is used to extract the three-axis attitude deflection sequence during the dwell phase to invert the environmental wind field vector, perform spatial vector superposition with the rotor wake envelope, delineate the windward undisturbed sampling area on the outer edge of the rotor wake envelope, track the azimuth plane of the windward undisturbed sampling area to generate a yaw compensation waypoint sequence, control the UAV to climb along the yaw compensation waypoint sequence, and simultaneously acquire atmospheric observation sequences and three-dimensional trajectory sequences. The hysteresis correction unit is used to perform time-domain alignment between the atmospheric observation sequence and the three-dimensional trajectory sequence, extract the sensor's response time constant, and invert the airflow disturbance rate by combining the instantaneous thrust distribution of the rotor in the windward undisturbed sampling area; the response time constant and the airflow disturbance rate are dynamically integrated to generate an altitude hysteresis compensation parameter, which is then injected into the vertical axis of the three-dimensional trajectory sequence to reconstruct the hysteresis-corrected altitude sequence. The profile generation unit performs vertical difference on the atmospheric observation sequence that matches the hysteresis-corrected height sequence, and extracts the vertical gradient vector that characterizes the stratification features. In the extreme value abrupt interval of the vertical gradient vector, the original sampling density of meteorological nodes is preserved, and in the smooth interval, feature aggregation and dimensionality reduction are performed on adjacent nodes to output the atmospheric vertical profile sequence with non-equidistant sampling.

Images