Unmanned aerial vehicle atmospheric boundary layer environment autonomous detection system and method

Abstract

The invention belongs to the technical field of boundary layer detection, and particularly relates to an unmanned aerial vehicle atmospheric boundary layer environment autonomous detection system and method.The unmanned aerial vehicle atmospheric boundary layer environment autonomous detection system comprises an unmanned aerial vehicle platform, a detection load subsystem, a flight control subsystem, a data processing subsystem and a ground station subsystem, and the unmanned aerial vehicle platform comprises an unmanned aerial vehicle, a flight control module and a communication module; the unmanned aerial vehicle is made of light-weight composite materials and has good structural strength and weather resistance, an equipment cabin, a battery cabin and a load cabin are arranged in a vehicle body, modular design is adopted, maintenance and upgrading are convenient, in addition, the unmanned aerial vehicle is further provided with an energy module and a power module, and energy and power are provided for the unmanned aerial vehicle; the flight control module comprises a flight control computer, an inertial measurement unit I MU and a GPS / Beidou dual-mode receiver. The unmanned aerial vehicle can effectively avoid movable obstacles such as birds in an atmospheric boundary layer and a strong turbulence area when carrying out environment detection, and the flight safety of the unmanned aerial vehicle during detection is guaranteed.

Description

Technical Field

[0001] The present invention belongs to the technical field of boundary layer detection, and in particular relates to an unmanned aerial vehicle (UAV) autonomous detection system and method for the atmospheric boundary layer environment. Background Art

[0002] The atmospheric boundary layer refers to the atmosphere directly affected by surface friction and thermal effects above the Earth's surface, and its thickness usually varies between several hundred meters and 2 kilometers. The detection of the atmospheric boundary layer environment can help people deeply understand the diffusion of pollutants, heat, and humidity exchange in the urban environment, help urban planners design more effective air circulation systems, reduce the accumulation of pollutants, and improve the air quality of cities. Traditional atmospheric boundary layer detection mainly relies on means such as meteorological towers, tethered balloons, sounding balloons, and remote sensing equipment, but these methods generally have disadvantages such as high costs.

[0003] With the development of UAV technology, the application of UAVs in the field of atmospheric detection has become increasingly widespread. For example, the patent with the publication number CN106125755B discloses an autonomous detection system and method for the atmospheric boundary layer environment of a UAV, including a main controller, an attitude sensor, a wireless communication module, a UAV flight control system, an LCD display, temperature, humidity, and air pressure sensors, a carbon dioxide sensor, and an SD card. The main controller controls the M100 aircraft to perform flight actions at a preset speed and angle by calling a series of SDK functions. At the same time, during the climbing process of the aircraft, the controller collects data at regular intervals or at a fixed altitude until a complete set of 0 - 500m profile measurements are completed, and then the aircraft returns and lands. It has the characteristics of low observation cost, flexible expandability of sensor types, simple and convenient operation, high data accuracy and reliability.

[0004] The above-mentioned existing devices also have the following deficiencies:

[0005] 1. In the above-mentioned existing device, the UAV needs to rely on artificially preset flight parameters and flight paths for environmental detection, and the preset flight route can only avoid known fixed obstacles in the detection environment. For birds or other moving obstacles and unknown strong turbulence areas existing in the atmospheric boundary layer, the above-mentioned existing device cannot avoid them, which is likely to cause accidents and thus affect the normal progress of environmental detection.

[0006] 2. When the above-mentioned existing technology is used for environmental pollution detection, it is necessary to conduct a full-scale detection of the target detection area, and then determine the location of the pollution source according to the particulate matter concentration distribution gradient. The determination efficiency of the pollution source location is relatively slow, and anti-diffusion measures cannot be taken in time for the pollution source, resulting in further spread of pollution. Summary of the Invention

[0007] The object of the present invention is to provide an autonomous detection system and method for the atmospheric boundary layer environment of an unmanned aerial vehicle in view of the problems raised in the above-mentioned background technology.

[0008] To achieve the above object, the present invention adopts the following technical solutions: An autonomous detection system for the atmospheric boundary layer environment of an unmanned aerial vehicle, comprising:

[0009] An unmanned aerial vehicle platform, the unmanned aerial vehicle platform includes an unmanned aerial vehicle, a flight control module and a communication module, the flight control module includes a flight control computer, an inertial measurement unit IMU and a GPS / Beidou dual-mode receiver, the flight control computer adopts a redundant design, and the communication module includes a radio data link, a 5G mobile communication terminal and a satellite communication terminal;

[0010] A detection payload subsystem for measuring environmental parameters of the atmospheric boundary layer;

[0011] A flight control subsystem that generates an optimal flight path according to the detection task requirements and real-time environmental parameter information for the navigation, guidance and control of the unmanned aerial vehicle;

[0012] A data processing subsystem that is responsible for the acquisition, preprocessing, quality control and preliminary data analysis of the detected environmental parameters;

[0013] A ground station subsystem for providing mission planning, real-time monitoring, data reception and in-depth data analysis.

[0014] Furthermore, the detection payload subsystem includes a meteorological sensor, a laser sensor and a millimeter-wave radar. The meteorological sensor is used to detect real-time environmental parameters, and the environmental parameters include temperature, humidity, wind speed, wind direction and turbulence intensity. The laser sensor is used to detect the real-time particulate matter concentration, and the millimeter-wave radar is used to obtain obstacle information.

[0015] Furthermore, the flight control subsystem includes a mission planning module, a meteorological data assimilation module, a path optimization module and a real-time adjustment module. The mission planning module is used to receive the detection task instructions sent by the ground station subsystem and generate a preliminary flight plan. The meteorological data assimilation module is used to fuse environmental parameters and construct a four-dimensional meteorological field that combines three-dimensional space and time. The path optimization module generates an optimal detection path based on the four-dimensional meteorological field information and the performance constraints of the unmanned aerial vehicle. The real-time adjustment module dynamically adjusts the flight path according to the real-time detection data and environmental changes obtained during the flight.

[0016] Furthermore, the processing flow of the data processing subsystem includes the following steps;

[0017] S1. Synchronously collect data from each sensor. The acquisition system adopts a distributed architecture, and each sensor node is connected to the main control computer through a CAN bus or Ethernet.

[0018] S2. Perform multi-source data synchronization through hardware timestamps, and use the Extended Kalman Filter (EKF) to fuse sensor data to eliminate the misjudgment of turbulence caused by the urban canopy effect, improving the measurement accuracy and reliability.

[0019] S3. Calculate the key parameters of the boundary layer, including the boundary layer height, the characteristics of the inversion layer, the turbulence intensity, the flux parameters, and the vertical distribution of aerosols.

[0020] S4. Use a lossy / lossless hybrid compression algorithm to reduce the data volume, and transmit the key data to the ground station subsystem through an adaptive communication link.

[0021] The data processing subsystem adopts an edge computing architecture to complete most of the computing tasks at the UAV end, reducing the communication burden and improving the system response speed.

[0022] Furthermore, the ground station subsystem includes a mission planning module, a flight monitoring module, a data reception and storage module, and a data analysis module. The mission planning module is used to lock the target detection area and select detection parameters according to the data of the input detection area. The flight monitoring module displays the UAV status information in real time. The data reception and storage module is used to manage the data from the UAV, including real-time telemetry data, regular batch data, emergency event reports, and system log information. The data analysis module is used to perform professional analysis on the data, including data visualization, spatio-temporal analysis, statistical analysis, model generalization, and report generation.

[0023] Furthermore, the path optimization module generates an optimal detection path based on meteorological field information and UAV performance constraints, including the following steps:

[0024] S1. Construct a four-dimensional dynamic network model in combination with the four-dimensional meteorological field, and update the wind field, the particulate matter concentration gradient, and the obstacle distribution in real time.

[0025] S2. Generate an initial path based on the four-dimensional dynamic network model using an improved A* algorithm, avoiding obstacle areas and no-fly zones, and optimize the path in real time according to the real-time data of the four-dimensional dynamic network model. The cost function comprehensively considers energy efficiency, safety constraints, and measurement coverage requirements. The real-time path optimization calculation method is as follows:

[0026] J = α·E L + β·S L + γ·C L

[0027] In the formula, E L represents the energy term, S LIndicates a safety item, C L Indicates an overlay item, and α, β, and γ represent the weight coefficients of each item respectively;

[0028] The calculation formula for the energy term is

[0029] E = 0.5·C d ·ρ air ·A·V rel 2

[0030] In the formula, C d is the drag coefficient, ρ air is the air density, and A is the frontal area of the UAV;

[0031] The safety item is implemented through the deep reinforcement learning DRL algorithm, and the safety item includes the following constraints:

[0032] Obstacle distance: The obstacle is detected in real time by a millimeter-wave radar, and the safety distance d > 10m is constrained. When obstacle avoidance is triggered, the deep reinforcement learning DRL local path replanning is started;

[0033] Turbulence intensity: The turbulent kinetic energy TKE is calculated based on the Reynolds stress tensor, and TKE < 0.5m 2 / s 3 is restricted within the flight area. When obstacle avoidance is triggered, the deep reinforcement learning DRL local path replanning is started;

[0034] The overlay item uses an improved A* algorithm to generate a global path covering the hot spot area, and the local measurement blind area is complemented in real time through incremental search RRT*;

[0035] The calculation formula for the optimal heading angle is:

[0036] θ opt = argminθ(α·E L (θ)+β·S L +γ·C L )

[0037] In the formula, θ opt represents the optimal heading angle.

[0038] Furthermore, the real-time adjustment module dynamically adjusts the flight path according to the real-time detection data and environmental changes obtained during the flight, including the following steps:

[0039] S1. Preset the particulate matter concentration threshold range, which includes the normal value range, the abnormal value range, and the dangerous value range;

[0040] S2. Based on the particulate matter concentration value collected by the laser sensor, determine the interval to which the current particulate matter concentration belongs. When the current particulate matter concentration is within the normal value interval, fly according to the initial path. When the current particulate matter concentration is in the abnormal value interval for three consecutive times, trigger the upwind tracking protocol to track the pollution source. When the current particulate matter concentration is in the dangerous value interval for three consecutive times, trigger the spiral search protocol to accurately locate the pollution source and determine the pollution source diffusion range;

[0041] S3. Generate an electronic mark of the pollution source, including coordinates, particulate matter concentration, and diffusion range, generate a detection report and send it to the ground station subsystem.

[0042] Furthermore, the upwind tracking protocol includes the following steps:

[0043] S1. Synchronously obtain particulate matter concentration, three-dimensional wind speed u / v / w components, and temperature and humidity data at a frequency of 10 Hz;

[0044] S2. Eliminate sensor noise through moving average filtering, with the window size N = 10;

[0045] S3. Establish a Gaussian plume model:

[0046] C(x,y,z) = Q / (2πuσ y σ z )·exp[-y 2 / (2σ y 2 )]·exp[-z 2 / (2σ z 2 )]

[0047] In the formula, u, v, and w respectively represent the components of the wind speed along the x, y, and z axes;

[0048] S4. Dynamically update the diffusion parameters and calculate the concentration gradient vector:

[0049] p_{k + 1} = p_k + α·▽f(p_k)

[0050] In the formula, α is the adaptive step size, and ▽f(p_k) is the concentration gradient estimated value;

[0051] S5. Real-time fuse wind speed and wind direction data to correct the moving direction and generate an upwind course angle:

[0052] θ_corrected = atan2(V_wind.y, V_wind.x) + π ± Δθ

[0053] where Δθ is dynamically fine-tuned according to the concentration gradient.

[0054] Furthermore, the spiral search protocol includes the following steps:

[0055] S1. With the current point as the center, expand according to the Archimedes spiral, preset the reference radius r0, the expansion coefficient α, the sampling interval Δθ = 30°, and r0 is adjusted according to the current wind speed:

[0056] r0 = 5 + 0.3·wind_speed;

[0057] S2. Establish a three-dimensional concentration field and perform nonlinear fitting using the Levenberg-Marquardt algorithm, specifically as follows

[0058] C(x,y) = A·exp[-((x - x0) 2 / 2σ x 2 +(y - y0) 2 / 2σ y 2 )]

[0059] S3. Determine whether it is a pollution source based on whether three consecutive sampling points satisfy the following two verification formulas. The determination formulas are as follows:

[0060] C n >0.9·C n-1

[0061] |C n -C n-1 |<5%·C n-1

[0062] S4. If it is determined to be a pollution source, automatically switch the altitude layer according to the vertical distribution of the concentration, and use fuzzy PID control to keep the altitude error < ±0.3m. Otherwise, continue to execute S1 - S3;

[0063] S5. When the semi-major axis length of the positioning error ellipse < 5m and the magnitude of the concentration gradient ‖▽C‖ < 1μg / m 3 / m, terminate the spiral search protocol.

[0064] The present invention also provides a method applied to the above-mentioned unmanned aerial vehicle atmospheric boundary layer environment autonomous detection system, including the following steps:

[0065] S1. The ground station subsystem receives the user's detection requirements, locks the target detection area and detection parameters;

[0066] S to 5M height and hovers, initialize each sensor, then each sensor starts to work, detect the environmental parameters, generate a preliminary flight path according to the environmental parameters, and then the unmanned aerial vehicle climbs to the detection starting height according to the preliminary flight path, and optimize the flight path in real time according to the environmental parameters during the climbing process;

[0067] S3. Perform environmental detection according to the optimized path, process the detection data in real time, transmit key data to the ground station subsystem, and receive ground station instructions for update;

[0068] S4. Start the return flight according to the plan or instructions and perform a safe landing;

[0069] S5. The ground station receives the complete data set, conducts professional analysis and visualization of the data, and generates a detection report.

[0070] Compared with the existing technologies, the advantages of the present invention are as follows:

[0071] 1. In the present invention, the flight path of the unmanned aerial vehicle is optimized in real time with the change of environmental parameters, and when optimizing the path, the energy consumption, flight safety and coverage are comprehensively considered. It can effectively avoid moving obstacles such as birds and strong turbulence areas within the atmospheric boundary layer, and ensure the flight safety during the detection of the unmanned aerial vehicle.

[0072] 2. When the present invention conducts environmental pollution detection, it dynamically adjusts the detection mode according to the particulate matter concentration in the detection environment. When the detected particulate matter concentration is in the abnormal value range, it automatically triggers the tracking protocol to track the pollution source. When the particulate matter concentration reaches the dangerous value range, it triggers the spiral search protocol to accurately locate the pollution source and determine the pollution diffusion range. It can quickly locate the pollution source, so that anti-diffusion measures can be taken earlier to avoid further pollution diffusion. Description of the Drawings

[0073] Figure 1 is a flowchart of a method applied to an unmanned aerial vehicle atmospheric boundary layer environment autonomous detection system provided by the present invention;

[0074] Figure 2 is a flowchart of the data processing subsystem in an unmanned aerial vehicle atmospheric boundary layer environment autonomous detection system provided by the present invention;

[0075] Figure 3 is a flowchart of an unmanned aerial vehicle atmospheric boundary layer environment autonomous detection system and a spiral search protocol provided by the present invention. Detailed Embodiments

[0076] The following embodiments are only for illustrative purposes and are not intended to limit the scope of the present invention.

[0077] Embodiment 1

[0078] As Figure 1 - Figure 2As shown in the figure, an autonomous detection system for the atmospheric boundary layer of an unmanned aerial vehicle (UAV) includes a UAV platform, a detection payload subsystem, a flight control subsystem, a data processing subsystem, and a ground station subsystem. The UAV platform includes a UAV, a flight control module, and a communication module. The UAV is made of lightweight composite materials, has good structural strength and weather resistance, and is equipped with an equipment cabin, a battery cabin, and a payload cabin inside the fuselage. The modular design facilitates maintenance and upgrade. In addition, the UAV is also equipped with an energy module and a power module to provide energy and power for the UAV. The flight control module includes a flight control computer, an inertial measurement unit (IMU), and a GPS / Beidou dual-mode receiver. The flight control computer adopts a redundant design to ensure the reliability of the system. The communication module includes a radio data link, a 5G mobile communication terminal, and a satellite communication terminal to achieve multi-mode communication guarantee;

[0079] The detection payload subsystem is used for measuring the environmental parameters of the atmospheric boundary layer. The detection payload subsystem includes a meteorological sensor, a laser sensor, and a millimeter-wave radar. The meteorological sensor is used to detect real-time environmental parameters, and the environmental parameters include temperature, humidity, wind speed, wind direction, and turbulence intensity. The laser sensor is used to detect the real-time particulate matter concentration, and the millimeter-wave radar is used to obtain obstacle information. The detection payload adopts a distributed layout design, and the main sensors are installed in the special nose fairing to ensure good air flow contact and reduce the interference of the UAV itself. Each sensor is connected to the data acquisition system through a standard interface, and the sampling frequency can be configured according to the detection requirements, usually 1-10 Hz;

[0080] The flight control subsystem generates an optimal flight path according to the detection task requirements and real-time environmental parameter information for the navigation, guidance, and control of the UAV. The flight control subsystem includes a mission planning module, a meteorological data assimilation module, a path optimization module, and a real-time adjustment module. The mission planning module is used to receive the detection task instructions sent by the ground station subsystem and generate a preliminary flight plan. The meteorological data assimilation module is used to fuse environmental parameters and construct a four-dimensional meteorological field that combines three-dimensional space and time. The path optimization module generates an optimal detection path based on the four-dimensional meteorological field information and the UAV performance constraints. The real-time adjustment module dynamically adjusts the flight path according to the real-time detection data and environmental changes obtained during the flight;

[0081] The path optimization module generates an optimal detection path based on the meteorological field information and the UAV performance constraints, including the following steps:

[0082] S1. Combine the four-dimensional meteorological field and adopt the WRF-LES coupling framework to construct a four-dimensional dynamic network model, and update the wind field, particulate matter concentration gradient, and obstacle distribution in real time;

[0083] S2. Combine with the four-dimensional dynamic network model, generate an initial path based on the improved A* algorithm, avoid obstacle areas and no-fly zones, and optimize the path in real time according to the real-time data of the four-dimensional dynamic network model. The cost function comprehensively considers energy efficiency, safety constraints, and measurement coverage requirements. The real-time path optimization calculation method is as follows:

[0084] J = α·E L +β·S L +γ·C L

[0085] In the formula, E L represents the energy term, S L represents the safety term, C L represents the coverage term, and α, β, and γ respectively represent the weight coefficients of each term;

[0086] The calculation formula for the energy term is

[0087] E = 0.5·C d ·ρ air ·A·V rel 2

[0088] In the formula, C d is the drag coefficient, ρ air is the air density, and A is the frontal area of the UAV;

[0089] The safety term is implemented through the deep reinforcement learning DRL algorithm, and the safety term includes the following constraints:

[0090] Obstacle distance: The obstacle is detected in real time by a millimeter-wave radar, and the safety distance d > 10m is constrained. When obstacle avoidance is triggered, the deep reinforcement learning DRL local path replanning is started;

[0091] Turbulence intensity: The turbulent kinetic energy TKE is calculated based on the Reynolds stress tensor, and TKE < 0.5m 2 / s 3 is restricted within the flight area. When obstacle avoidance is triggered, the deep reinforcement learning DRL local path replanning is started;

[0092] The coverage term uses the improved A* algorithm to generate a global path covering the hot spot area, and the local measurement blind area is complemented in real time through incremental search RRT*;

[0093] The calculation formula for the optimal heading angle is:

[0094] θ opt = argminθ(α·E L (θ)+β·S L +γ·C L )

[0095] In the formula, θ optIndicates the optimal heading angle;

[0096] A data processing subsystem, which is responsible for the acquisition, preprocessing, quality control, and preliminary data analysis of environmental parameter detection. The processing flow of the data processing subsystem includes the following steps;

[0097] S1. Synchronously collect data from each sensor. The acquisition system adopts a distributed architecture, and each sensor node is connected to the main control computer through a CAN bus or Ethernet;

[0098] S2. Perform multi-source data synchronization through hardware timestamps, and use the Extended Kalman Filter (EKF) to fuse sensor data to eliminate the misjudgment of turbulence caused by the urban canopy effect and improve the measurement accuracy and reliability;

[0099] S3. Calculate the key parameters of the boundary layer, including the boundary layer height, inversion layer characteristics, turbulence intensity, flux parameters, and aerosol vertical distribution;

[0100] The specific calculation method is as follows:

[0101] The boundary layer height is calculated using the standard deviation of the turbulent pulsation velocity σ_w threshold determination. The threshold is set to 0.3M. Based on the turbulent kinetic energy (TKE) profile output by the WRF model, the boundary layer top is determined by the threshold method;

[0102] The inversion layer characteristics are updated using the four-dimensional variational assimilation (4D-Var) technology for the spatio-temporal evolution of the inversion layer;

[0103] The formula for turbulence intensity is: TKE = 0.5·(u 2 + v 2 + w 2 ), where u, v, and w represent the components of the wind speed along the x, y, and z axes respectively;

[0104] The flux parameters are calculated by the eddy covariance method, specifically: H = ρ·Cp·w'θ', where ρ is the air density and Cp is the specific heat at constant pressure;

[0105] The aerosol vertical distribution is solved using the Fernald method for aerosol optical depth;

[0106] S4. Use a lossy / lossless hybrid compression algorithm to reduce the data volume and transmit the key data to the ground station subsystem through an adaptive communication link;

[0107] The data processing subsystem adopts an edge computing architecture to complete most of the computing tasks at the UAV end, reducing the communication burden and improving the system response speed;

[0108] The ground station subsystem is used to provide mission planning, real-time monitoring, data reception, and in-depth data analysis. The ground station subsystem includes a mission planning module, a flight monitoring module, a data reception and storage module, and a data analysis module. The mission planning module is used to lock the target detection area and select detection parameters according to the data of the input detection area. The flight monitoring module displays the UAV status information in real time. The data reception and storage module is used to manage the data from the UAV, including real-time telemetry data, regular batch data, emergency event reports, and system log information. The data analysis module is used to perform professional analysis on the data, including data visualization, spatio-temporal analysis, statistical analysis, model generalization, and report generation. The software of the ground station subsystem adopts a microservices architecture, and each functional module can be independently deployed and expanded. The system supports data exchange with external meteorological business systems, facilitating the integration of detection data into the business forecasting process.

[0109] Embodiment 2

[0110] As Figure 3 shown, the real-time adjustment module dynamically adjusts the flight path according to the real-time detection data and environmental changes obtained during the flight, including the following steps:

[0111] S1. Preset the particulate matter concentration threshold range, which includes a normal value range, an abnormal value range, and a dangerous value range;

[0112] S2. According to the particulate matter concentration value collected by the laser sensor, determine the interval to which the current particulate matter concentration belongs. When the current particulate matter concentration is in the normal value range, fly according to the initial path. When the current particulate matter concentration is in the abnormal value range for three consecutive times, trigger the upwind tracking protocol to track the pollution source. When the current particulate matter concentration is in the dangerous value range for three consecutive times, trigger the spiral search protocol to accurately locate the pollution source and determine the pollution source diffusion range;

[0113] S3. Generate an electronic mark for the pollution source, including coordinates, particulate matter concentration, and diffusion range, generate a detection report, and send it to the ground station subsystem.

[0114] The upwind tracking protocol includes the following steps:

[0115] S1. Synchronously obtain particulate matter concentration, three-dimensional wind speed u / v / w components, temperature and humidity data at a frequency of 10Hz;

[0116] S2. Eliminate sensor noise through moving average filtering, and the window size N = 10;

[0117] S3. Establish a Gaussian plume model:

[0118] C(x,y,z) = Q / (2πuσ y σ z )·exp[-y 2 / (2σ y 2 )]·exp[-z 2 / (2σ z 2 )]

[0119] In the formula, u, v, and w respectively represent the components of the wind speed along the x, y, and z axes;

[0120] S4. Dynamically update the diffusion parameters and calculate the concentration gradient vector:

[0121] p_{k + 1}=p_k + α·▽f(p_k)

[0122] In the formula, α is the adaptive step size, and ▽f(p_k) is the estimated value of the concentration gradient;

[0123] S5. Real-time fuse the wind speed and wind direction data to correct the moving direction and generate the upwind course angle:

[0124] θ_corrected = atan2(V_wind.y, V_wind.x)+π ± Δθ

[0125] where Δθ is dynamically fine-tuned according to the concentration gradient;

[0126] The spiral search protocol includes the following steps:

[0127] S1. Expand in an Archimedean spiral with the current point as the center, preset the reference radius r0, the expansion coefficient α, and the sampling interval Δθ = 30°, and r0 is adjusted according to the current wind speed:

[0128] r0 = 5 + 0.3·wind_speed;

[0129] S2. Establish a three-dimensional concentration field and perform nonlinear fitting using the Levenberg-Marquardt algorithm, specifically as follows

[0130] C(x,y)=A·exp[-((x - x0) 2 / 2σ x 2 +(y - y0) 2 / 2σ y 2 )]

[0131] S3. Determine whether it is a pollution source by whether the following two verification formulas are satisfied for three consecutive sampling points. The determination formulas are as follows:

[0132] C n >0.9·C n-1

[0133] |C n -Cn-1 | < 5%·C n-1

[0134] S4. If it is determined as a pollution source, automatically switch the altitude layer according to the vertical distribution of the concentration, and use fuzzy PID control to keep the altitude error < ±0.3 m. Otherwise, continue to execute S1 - S3;

[0135] S5. When the semi - major axis length of the positioning error ellipse < 5 m and the magnitude of the concentration gradient ‖▽C‖ < 1 μg / m 3 / m, terminate the spiral search protocol.

[0136] The present invention also provides a method applied to the above - mentioned unmanned aerial vehicle atmospheric boundary layer environment autonomous detection system, including the following steps:

[0137] S1. The ground station subsystem receives the user's detection requirements, locks the target detection area and detection parameters;

[0138] S2. The unmanned aerial vehicle takes off to a height of 5 m and hovers, initializes each sensor, and then each sensor starts to work, detects the environmental parameters, generates a preliminary flight path according to the environmental parameters, and then the unmanned aerial vehicle climbs to the detection starting height according to the preliminary flight path, and optimizes the flight path in real - time according to the environmental parameters during the climbing process;

[0139] S3. Conduct environmental detection according to the optimized path, process the detection data in real - time and transmit the key data to the ground station subsystem, and receive the ground station instructions for update;

[0140] S4. Start to return according to the plan or instructions and execute a safe landing;

[0141] S5. The ground station receives the complete data set, conducts professional analysis and visualization on the data, and generates a detection report.

[0142] The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. An autonomous detection system for the atmospheric boundary layer environment of an unmanned aerial vehicle, characterized in that, Including: A drone platform, which includes a drone, a flight control module, and a communication module. The flight control module includes a flight control computer, an inertial measurement unit (IMU), and a GPS / Beidou dual-mode receiver. The flight control computer adopts a redundant design. The communication module includes a radio data link, a 5G mobile communication terminal, and a satellite communication terminal; A detection payload subsystem for measuring environmental parameters at the boundary of the atmosphere; A flight control subsystem that generates an optimal flight path according to the detection task requirements and real-time environmental parameter information for the navigation, guidance, and control of the drone; A data processing subsystem responsible for the acquisition, preprocessing, quality control, and preliminary data analysis of the detected environmental parameters; A ground station subsystem for providing mission planning, real-time monitoring, data reception, and in-depth data analysis.

2. The autonomous detection system for the atmospheric boundary layer environment of an unmanned aerial vehicle according to claim 1, characterized in that The detection payload subsystem includes a meteorological sensor, a laser sensor, and a millimeter-wave radar. The meteorological sensor is used to detect real-time environmental parameters, which include temperature, humidity, wind speed, wind direction, and turbulence intensity. The laser sensor is used to detect the real-time particulate matter concentration, and the millimeter-wave radar is used to obtain obstacle information.

3. The autonomous detection system for the atmospheric boundary layer environment of an unmanned aerial vehicle according to claim 1, wherein The flight control subsystem includes a mission planning module, a meteorological data assimilation module, a path optimization module, and a real-time adjustment module. The mission planning module is used to receive the detection task instructions sent by the ground station subsystem and generate a preliminary flight plan. The meteorological data assimilation module is used to fuse environmental parameters and construct a four-dimensional meteorological field that integrates three-dimensional space and time. The path optimization module generates an optimal detection path based on the four-dimensional meteorological field information and the performance constraints of the drone. The real-time adjustment module dynamically adjusts the flight path according to the real-time detection data and environmental changes obtained during the flight.

4. The autonomous detection system for the atmospheric boundary layer environment of an unmanned aerial vehicle according to claim 1, characterized in that, The processing flow of the data processing subsystem includes the following steps; S1. Synchronously collect data from each sensor. The acquisition system adopts a distributed architecture, and each sensor node is connected to the main control computer through a CAN bus or Ethernet; S2. Perform multi-source data synchronization through a hardware timestamp, and use an extended Kalman filter (EKF) to fuse sensor data to eliminate the misjudgment of turbulence caused by the urban canopy effect and improve the measurement accuracy and reliability; S3. Calculate the key parameters of the boundary layer, including the boundary layer height, the characteristics of the inversion layer, the turbulence intensity, the flux parameters, and the vertical distribution of aerosols; S4. Use a lossy / lossless hybrid compression algorithm to reduce the data volume and transmit the key data to the ground station subsystem through an adaptive communication link.

5. The autonomous detection system for the atmospheric boundary layer environment of an unmanned aerial vehicle according to claim 1, characterized in that, The ground station subsystem includes a mission planning module, a flight monitoring module, a data reception and storage module, and a data analysis module. The mission planning module is used to lock the target detection area and select detection parameters according to the data of the input detection area. The flight monitoring module displays the UAV status information in real time. The data reception and storage module is used to manage the data from the UAV, including real-time telemetry data, periodic batch data, emergency event reports, and system log information. The data analysis module is used to perform professional analysis on the data, including data visualization, spatio-temporal analysis, statistical analysis, model generalization, and report generation.

6. The autonomous detection system for the atmospheric boundary layer environment of a drone according to claim 3, characterized in that [[ID= ​ ​ J = α·E L + β·S L + γ·C L where, E L represents the energy term, S L represents the safety term, C L represents the coverage term, and α, β, and γ respectively represent the weight coefficients of each term; ​ E = 0.5·C d ·ρ air ·A·V rel 2 where C d is the drag coefficient, ρ air is the air density, and A is the frontal area of the UAV; ​ ​ Turbulence intensity: Calculate the turbulent kinetic energy TKE based on the Reynolds stress tensor, and limit TKE < 0.5 m 2 / s 3 , and start the deep reinforcement learning DRL local path replanning when triggering obstacle avoidance; ​ ​ θ opt = argminθ(α·E L (θ)+β·S L +γ·C L ) where θ opt represents the optimal course angle.

7. The autonomous detection system for the atmospheric boundary layer environment of an unmanned aerial vehicle according to claim 3, wherein, ​ ​ ​ ​ 8. An unmanned aerial vehicle atmospheric boundary layer environment autonomous detection system according to claim 7, characterized in that, ​ ​ ​ ​ C(x,y,z) = Q / (2πuσ y σ z )·exp[-y 2 / (2σ y 2 )]·exp[-z 2 / (2σ z 2 )] ​ ​ where α is the adaptive step size, is the estimated value of the concentration gradient; ​ θ_corrected = atan2(V_wind.y, V_wind.x) + π ± Δθ where Δθ is dynamically fine-tuned according to the concentration gradient.

9. The autonomous detection system for the atmospheric boundary layer environment of an unmanned aerial vehicle according to claim 7, characterized in that, The spiral search protocol includes the following steps: S1. With the current point as the center, expand according to the Archimedean spiral, preset the reference radius r0, the expansion coefficient α, the sampling interval Δθ = 30°, and r0 is adjusted according to the current wind speed: r0 = 5 + 0.3 · wind_speed; S2. Establish a three-dimensional concentration field and perform nonlinear fitting using the Levenberg-Marquardt algorithm as follows C(x,y) = A·exp[-((x - x0) 2 / 2σ x 2 +(y - y0) 2 / 2σ y 2 )] S3. Determine whether it is a pollution source based on whether the following two verification formulas are satisfied by three consecutive sampling points. The determination formulas are as follows: C n > 0.9·C n-1 |C n -C n-1 |<5%·C n-1 S4. If it is determined to be a pollution source, automatically switch the altitude layer according to the vertical distribution of the concentration, and use fuzzy PID control to keep the altitude error < ±0.3m. Otherwise, continue to execute S1 - S3; S5. When the semi-major axis length of the positioning error ellipse is < 5 m and the modulus of the concentration gradient terminate the spiral search protocol.

10. A method applied to the unmanned aerial vehicle atmospheric boundary layer environment autonomous detection system described in claims 1-9, characterized in that, including the following steps: S1. The ground station subsystem receives the user's detection requirements, locks the target detection area and detection parameters; S2. The UAV takes off and hovers at a height of 5M, initializes each sensor, and then each sensor starts to work, detects the environmental parameters, generates a preliminary flight path according to the environmental parameters, and then the UAV climbs to the detection starting height according to the preliminary flight path. During the climbing process, the flight path is optimized in real time according to the environmental parameters; S3. Conduct environmental detection according to the optimized path, process the detection data in real time and transmit the key data to the ground station subsystem, and receive the ground station instructions for update; S4. Start to return according to the plan or instruction and perform a safe landing; S5. The ground station receives the complete data set, conducts professional analysis and visualization of the data, and generates a detection report.

Images