Environment monitoring method based on low-altitude aircraft

Abstract

The invention discloses an environment monitoring method based on a low-altitude aircraft, belongs to the technical field of low-altitude aircrafts, and realizes high-precision monitoring through cooperation of anti-interference hardware, edge calculation, navigation communication and cloud modeling. According to the system, a hexagonal prism cabin body is integrated with multiple types of sensors, and data are dynamically calibrated through a BP neural network. And aligning the multi-source spatio-temporal data by using extended Kalman filtering, and generating a three-dimensional grid model containing obstacles and a flow field. Navigation adopts Beidou / IMU combination and visual SLAM redundancy switching, and communication is guaranteed through a 5G private network and UWB double links. And the cloud constructs a three-dimensional AQI thermodynamic diagram based on inverse distance weighting and a Kriging algorithm, predicts pollutant diffusion in combination with a random forest, and drives the aircraft to execute a spiral / grid composite route. The problems that traditional monitoring is low in density, complex in model and insufficient in precision are solved, and a full-link anti-interference perception-decision closed loop is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of low-altitude aircraft, and in particular to an environmental monitoring method based on a low-altitude aircraft. Background Art

[0002] With the expanding application of low-altitude aircraft in environmental monitoring, urban governance, emergency rescue, and other fields, the demand for high-precision data collection and real-time decision-making in complex environments is becoming increasingly urgent. Traditional low-altitude monitoring systems rely on single sensors or simple sensor combinations. Under complex operating conditions such as high humidity, vibration, and electromagnetic interference, they face problems such as poor data reliability, low fusion accuracy, and insufficient decision support, making them difficult to meet the requirements of refined monitoring tasks. Consequently, higher technical requirements are being placed on aircraft in terms of sensor reliability, data fusion accuracy, navigation and communication stability, and cloud-based data processing capabilities.

[0003] To date, research on environmental monitoring technology based on low-altitude aircraft is relatively rare. A patent application by Chen Tao, Qiao Xuesong, and others, titled "Regional Environmental Monitoring System, Method, and Apparatus Based on Spatiotemporal Remote Sensing," focuses on optimizing the dynamic window range, combining multiple algorithms, and applying spatiotemporal data interpolation techniques to significantly improve the accuracy and completeness of regional environmental monitoring. This approach flexibly adapts to data variations under varying spatiotemporal conditions, achieving higher precision and more comprehensive coverage. However, this approach relies heavily on the quality of raw sensor data and lacks interoperability, significantly impacting data accuracy and interoperability across different data modalities. In their paper, "Research on the Synergy of Micropressure Sensor Sensitivity and Overload Impedance and Dynamic Interference Decoupling," scholars Yu Zhongliang and others propose an innovative approach using a differential operating mechanism to effectively suppress vibration interference. This approach offers significant theoretical advantages, improving system stability and anti-interference capabilities. However, in practical applications, this research has yet to fully address the challenges posed by condensation in high-humidity environments.

[0004] Therefore, a low-altitude aircraft environmental monitoring technology is needed. Addressing the data reliability, fusion accuracy, and decision support issues faced by traditional systems under complex operating conditions, this technology can overcome the bottlenecks of existing technologies, such as their reliance on raw data quality, insufficient collaborative capabilities, and limited environmental adaptability. This will meet the demand for high-precision, real-time monitoring from low-altitude aircraft in fields such as environmental monitoring and urban governance, improve system performance and practicality, and provide technical support for related fields. Summary of the Invention

[0005] The main purpose of the present invention is to provide an environmental monitoring method based on a low-altitude aircraft, which can effectively solve the problems mentioned in the background technology.

[0006] To achieve the above object, the technical solution adopted by the present invention is: An environmental monitoring method based on a low-altitude aircraft comprises the following steps: S01. Sensor Cabin Construction: The sensor cabin for the low-altitude aircraft is designed as a hexagonal prism. The cabin surface is hard-anodized. A honeycomb support frame is constructed inside the cabin, and the frame is connected to the cabin frame using titanium alloy bolts. S02. Sensor Integration and Circuit Protection: Environmental sensors are installed on the six sides of the cabin. Each sensor is equipped with an independent PCB adapter board, which is connected to the main control board via a flexible flat cable. The PCB adapter board integrates ESD protection diodes and common-mode inductors. S03. Dynamic Real-Time Calibration: A dynamic calibration algorithm based on a BP neural network is constructed to compensate for sensor drift caused by temperature and humidity changes in real time. S04. Spatiotemporal Alignment and Modeling of Multi-Source Data: Build a timestamp mapping matrix for sensor data of different sampling frequencies. Use the extended Kalman filter algorithm to achieve spatiotemporal alignment of multi-source data, and generate a 3D environmental mesh model after fusion processing. S05. Navigation and communication redundancy design: This system utilizes navigation routing technology that integrates the Beidou satellite navigation system with an inertial measurement unit, automatically switching navigation modes based on signal quality. A dual-link redundant communication system is also constructed: the primary link utilizes a 5G private network to connect to mobile edge computing nodes, while the backup link utilizes a UWB ultra-wideband communication module. S06. Initial modeling of pollutant concentration fields: Processing discrete monitoring point data based on the inverse distance weighted algorithm: Determine the set of monitoring points involved in interpolation, calculate the Euclidean distance between the target point and each monitoring point, and generate an initial two-dimensional concentration field by weighted summation of the concentration values ​​at each monitoring point. S07. Concentration Field Optimization and Visualization: Optimize the initial concentration field using the Kriging algorithm. Analyze spatial variation characteristics using the variogram, determine semivariogram model parameters, calculate target point predictions based on the unbiased optimal estimation principle, and generate a three-dimensional AQI heat map. S08. Adaptive Route Planning: Identifies high-pollution areas based on a 3D AQI heat map and automatically plans complex routes: Spiral routes are used for refined monitoring in pollution hotspots, while grid routes are used to improve coverage efficiency in non-hotspot areas. S09. Pollutant Diffusion Prediction: After receiving regional monitoring data, the pollutant diffusion rate is predicted using a random forest algorithm based on historical pollution diffusion data and real-time meteorological conditions.

[0007] Preferably, the sensor cabin hardware architecture in S02 includes: S21. Install the temperature and humidity sensor, air pressure sensor, and ultrasonic anemometer on three sides of the cabin. Secure the sensors with aluminum alloy brackets, which are connected to the cabin frame with screws. S22. A micro differential absorption spectrometer and PM2.5 laser sensor are mounted on the same side. The spectrometer is bonded to the cabin shell via a thermally conductive silicone pad. A guide groove is designed around the laser sensor. An active air intake system is constructed using a micro air pump, and a waterproof and breathable membrane is installed at the air intake. S23. The lidar and multispectral camera are installed on the cabin roof, using an electronic level to assist with installation. All sensor cables are secured through cable management troughs inside the cabin, using shielded twisted-pair cables for signal transmission, and shielded metal connectors at the interfaces. S24. Sandblast the surface of the cabin and apply a super-hydrophobic coating using chemical vapor deposition technology. S25. Each sensor is mounted using four shock-absorbing pads arranged in a rectangular pattern.

[0008] Preferably, the dynamic real-time calibration of S03 is implemented by an edge computing processing unit, including: S31. Build a BP neural network model with three hidden layers. The first hidden layer has 64 neurons, the second hidden layer has 32 neurons, and the third hidden layer has 16 neurons. The input layer receives the original temperature and humidity signals and the temperature compensation coefficient, and the output layer outputs the calibration parameters. S32. Generate a standard environment using a high-precision temperature and humidity calibration chamber, and simultaneously collect the sensor's raw output and standard values ​​to construct a training dataset. S33. The Adam optimization algorithm is used to train the neural network for 500 cycles. The edge computing unit uses a 50ms cycle to input real-time data into the trained model and output calibration parameters.

[0009] Preferably, the specific steps of generating the three-dimensional environment grid model in S04 include: S41. Using the extended Kalman filter algorithm, the current state is predicted based on the state estimate at the previous moment in the prediction phase; S42. In the update phase, the actual measured value is integrated with the predicted value; S43. Repeat the iterative calculations of S41-S42 to make the timestamp alignment error ≤ ±10 μs; S44. Fuse the spatiotemporally aligned lidar point cloud with camera image data and extract obstacle contours using a point cloud segmentation algorithm. S45. Use the optical flow method to calculate the flow field vector, with a 10cm 3 The grid resolution generates a three-dimensional environment grid model including obstacle distribution and flow field characteristics.

[0010] Preferably, the navigation communication system implementation of S05 includes: S51. Use a combined navigation solution with a BeiDou-3 RTK receiver and a high-precision IMU. S52. Receive differential correction data from ground-based augmentation stations to achieve centimeter-level positioning with a data update rate of 10Hz; S53. Automatically switch to visual SLAM navigation mode when Beidou signal quality degrades; S54 uses the ORB-SLAM3 algorithm framework, equipped with an 8K camera, and extracts feature points at 500fps using FAST corner detection and BRIEF descriptors. S55. Combine IMU data with visual-inertial tight coupling to achieve monocular visual positioning accuracy of ±0.5m, and verify navigation data consistency through Kalman filtering; S56. The primary link uses a module that supports the 5G NR standard to connect to the mobile edge computing node, configured with a 20MHz bandwidth carrier, and an end-to-end latency of less than 10ms; S57. The backup link uses a UWB module (operating frequency band 3.1-10.6 GHz) with DS-UWB modulation technology, automatically switching if the primary link is interrupted. S58. Integrates a 2-18GHz broadband spectrum detection module that scans the electromagnetic environment with a 10ms cycle and triggers frequency band switching when the interference signal power is greater than -60dBm.

[0011] Compared with the prior art, the present invention has the following beneficial effects: This invention proposes an innovative intelligent monitoring system to solve the monitoring problems of low-altitude aircraft in complex environments. It adopts a hexagonal prismatic sensor cabin, integrates a high-precision detection unit, and significantly improves the accuracy and data rate of the sensor through special coatings and shock-absorbing structures; uses edge computing and advanced algorithms to greatly reduce sensor errors, and improves the alignment accuracy of multi-source data to generate a high-precision three-dimensional environmental model; the navigation system and redundant communication technology ensure the reliability of high-precision positioning and data transmission; the cloud platform constructs an accurate three-dimensional AQI heat map through advanced algorithms, and improves the efficiency of pollution monitoring through decision-making models. Overall, the system has achieved significant improvements in multiple technical indicators, effectively improving the performance of environmental monitoring of low-altitude aircraft. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 Construct a flow chart for the hardware architecture of the present invention; Figure 2 is a data processing flow chart of the present invention; Figure 3 This is a flow chart of the navigation communication system of the present invention. DETAILED DESCRIPTION

[0013] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0014] The present invention provides an environmental monitoring system and method based on a low-altitude aircraft, the steps of which are as follows: S01、 Figure 1 The hardware architecture construction flow chart shown in the figure shows that the low-altitude aircraft described in this invention uses a hexagonal prismatic sensor cabin as its core carrier. The cabin adopts a hexagonal prismatic structure and has a hard anodized surface. A honeycomb support skeleton is constructed inside the cabin, and the skeleton is connected to the cabin frame with titanium alloy bolts. S02, six side-mounted accessories for the cabin, with an independent PCB adapter board designed for each sensor, connected to the main control board via a flexible flat cable, with ESD electrostatic protection diodes and common-mode inductors integrated on the adapter board; S21. Install the temperature and humidity sensor, air pressure sensor, and ultrasonic anemometer on the three sides of the cabin. Use aluminum alloy brackets to fix the sensors. Connect the brackets to the cabin frame with screws to ensure they are firmly installed. For example, the installation of the cabin's side hardware adheres to the principles of functional zoning and anti-interference design: the temperature and humidity sensor, air pressure sensor, and ultrasonic anemometer are embedded in the center of the front, left, and right sides of the cabin, respectively. They are three-dimensionally mounted using L-shaped aluminum alloy brackets. The bottom of the bracket is fastened to the cabin frame with M3 titanium alloy screws and pre-set screw holes. The top uses elastic clips to nest the sensor body, leaving a 3mm expansion gap to accommodate thermal expansion and contraction. Each sensor is equipped with a separate PCB adapter board with a 1206 package ESD protection diode and an 80μH common-mode inductor. It is connected to the main control board via a 0.3mm pitch flexible flat cable, covered with a tinned copper braided shield, and terminated with a metal shield connector crimped to ground.

[0015] The S22 and micro differential absorption spectrometer PM2.5 laser sensors are mounted on the same side. The spectrometer is attached to the cabin shell via a thermally conductive silicone pad, using the shell for passive heat dissipation. A guide groove is designed around the laser sensor, and an active air intake system is constructed using a micro air pump. A waterproof and breathable membrane is installed at the air inlet to prevent liquid ingress while ensuring normal gas circulation. The S23, LiDAR, and multispectral camera are installed on the cabin roof, using an electronic level to assist in installation. All sensor cables are secured and organized in cable management troughs inside the cabin. Shielded twisted-pair cables are used for signal transmission, and metal shielded connectors are used at the interfaces. S24. Sandblast the surface of the cabin and apply a super-hydrophobic coating using chemical vapor deposition technology. S25. Use 4 shock-absorbing pads for each sensor, distributed in a rectangular arrangement. S03, dynamic calibration algorithm based on BP neural network, used to compensate for the measurement drift of the sensor caused by temperature and humidity changes in real time; S31, such as Figure 2 The data processing flow chart shown here builds a BP neural network model with three hidden layers, where the first hidden layer has 64 neurons, the second hidden layer has 32 neurons, and the third hidden layer has 16 neurons. The input layer receives the original signal from the temperature and humidity sensor and the temperature compensation coefficient, and the output layer outputs the calibrated temperature and humidity parameters. S32. Collect working condition data, generate a standard environment through a high-precision temperature and humidity calibration box, synchronously collect the original output of the sensor and the standard value, and build a training data set; S33. Use the Adam optimization algorithm to train the BP neural network for 500 training cycles. The edge computing unit uses a 50ms cycle to input the real-time collected temperature and humidity raw signals and temperature compensation coefficients into the trained BP neural network model, and outputs calibration parameters to correct the sensor data. For example, during the training phase, a constant temperature and humidity chamber is used to simulate environmental conditions, collecting 2,000 sets of sensor signals and their corresponding standard values. The data is then normalized using a Z-score and fed into the neural network for training. The sensors output signals at a frequency of 200 Hz. After interference is filtered out by an isolation amplifier circuit, the edge computing unit packages the temperature and humidity values ​​and compensation coefficients every 50 milliseconds and feeds them into the BP neural network model. The calibration parameters are then synchronously transmitted to the data acquisition module via the I2C bus. The corrected data is then stored synchronously with the meteorological data in a dual-port random access memory, forming a raw data set with a hardware timestamp. This data provides a high-frequency calibrated data source for the timestamp mapping matrix, ensuring the accuracy of the benchmark data for multi-sensor temporal alignment.

[0016] S04. For sensors with different sampling frequencies, such as lidar, cameras, and meteorological sensors, a timestamp mapping matrix is ​​established. By recording the acquisition time of each sensor data, a time-data index table is constructed to achieve the basis for data time dimension alignment; S41. Use the extended Kalman filter algorithm to predict and update the data of different sensors. In the prediction stage, the current state is predicted based on the state estimate value of the previous moment and the system model; S42, in the update phase, the actual measured values ​​are fused with the predicted values ​​to obtain a more accurate state estimate; S43. Repeat the operations of S41 and S42 to unify the time bases of the multi-source data through iterative calculation, and control the timestamp alignment error within ±10μs.

[0017] S44, fusing the spatiotemporally aligned lidar point cloud data with the camera image data, and extracting obstacle contours using a point cloud segmentation algorithm; S45. Calculate the flow field vector using the optical flow method, using a 10 cm × 10 cm × 10 cm grid resolution to generate a three-dimensional environment grid model that includes the spatial distribution of obstacles and flow field motion characteristics; S05, using integrated navigation route technology and dual-link redundant communication for design, built a navigation route technology system that integrates the Beidou satellite navigation system + inertial measurement unit combination, and automatically switches navigation modes according to the signal environment; S51, adopt the combined navigation solution of BeiDou satellite navigation system and inertial measurement unit, select RTK receiver supporting BeiDou-3 system, and cooperate with high-precision IMU; S52, by receiving the differential correction data from the ground-based augmentation station in real time, centimeter-level positioning is achieved, with a data update rate of 10Hz; S53. When the signal quality of the Beidou satellite navigation system deteriorates, the system automatically switches to the visual SLAM navigation mode; S54 uses the ORB-SLAM3 algorithm as the core processing framework, is equipped with an 8K resolution visual camera, and uses FAST corner detection and BRIEF descriptor to extract feature points. The feature point extraction speed can reach 500fps. S55, combines IMU data for tight visual-inertial coupling to achieve monocular visual positioning accuracy of ±0.5m. The system has built-in navigation mode switching judgment logic and uses Kalman filtering to perform consistency verification on the Beidou satellite navigation system and visual SLAM data; For example, the navigation mode transition logic involves a comprehensive assessment of signal quality, including monitoring the GNSS receiver's carrier signal-to-noise ratio and multipath error. When the C / N0 value for five consecutive cycles is below 35dBHz and the error exceeds 1.5 meters, a transition to visual simultaneous positioning and mapping mode is activated. The Kalman filter module uses Mahalanobis distance to evaluate GNSS and visual SLAM data. If data exceeds a set threshold, it excludes the anomalous data to ensure visual positioning accuracy remains within ±0.5 meters. A mode transition signal is then sent to the communication system to optimize data transmission.

[0018] S56, Main Link Communication: Uses 5G private network as the main data transmission channel, connected to the mobile edge computing node; S57. Select a communication module that supports the 5GNR standard and configure a 20MHz bandwidth carrier to transmit raw sensor data. The MEC node processes the data locally, keeping the end-to-end latency below 10ms. S58, deploy UWB ultra-wideband communication modules as backup links, operating in the 3.1-10.6 GHz frequency band and using DS-UWB modulation technology. When the primary link is interrupted due to signal obstruction, interference, etc., the system automatically and seamlessly switches to the backup link. S59, integrated 2-18GHz broadband spectrum detection module, real-time scanning of the electromagnetic environment with a period of 10ms. When the interference signal power is detected > -60dBm, the frequency band switching logic is triggered; S06. Process the discrete monitoring point data based on the inverse distance weighted algorithm, determine the set of monitoring points involved in interpolation, calculate the Euclidean distance between the target point and each monitoring point; perform weighted summation on the concentration values ​​of each monitoring point to generate an initial two-dimensional concentration field, and establish a prototype of the spatial distribution of pollutant concentrations; S07. Use the Kriging algorithm to optimize the initial concentration field, analyze the spatial variation characteristics of the monitoring data through the variogram, determine the semivariogram model parameters, calculate the predicted value of the target point based on the unbiased optimal estimation principle, and generate a three-dimensional AQI heat map; S08: Identify high-pollution areas based on the three-dimensional AQI heat map and automatically plan a "spiral + grid" composite flight path. In pollution hotspots, spiral flights are used for refined monitoring, while in non-hotspot areas, grid flights are used to improve coverage efficiency. S09. After receiving regional data, the model predicts the pollutant diffusion rate through the random forest algorithm based on historical pollution diffusion data and real-time meteorological conditions.

[0019] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A low-altitude aircraft-based environmental monitoring method, characterized by: The following steps are involved: S01. Sensor Cabin Construction: The sensor cabin for the low-altitude aircraft is designed as a hexagonal prism. The cabin surface is hard-anodized. A honeycomb support frame is constructed inside the cabin, and the frame is connected to the cabin frame using titanium alloy bolts. S02. Sensor Integration and Circuit Protection: Environmental sensors are installed on the six sides of the cabin. Each sensor is equipped with an independent PCB adapter board, which is connected to the main control board via a flexible flat cable. The PCB adapter board integrates ESD protection diodes and common-mode inductors. S03. Dynamic Real-Time Calibration: A dynamic calibration algorithm based on a BP neural network is constructed to compensate for sensor drift caused by temperature and humidity changes in real time. S04. Spatiotemporal Alignment and Modeling of Multi-Source Data: Build a timestamp mapping matrix for sensor data of different sampling frequencies. Use the extended Kalman filter algorithm to achieve spatiotemporal alignment of multi-source data, and generate a 3D environmental mesh model after fusion processing. S05. Navigation and communication redundancy design: This system utilizes navigation routing technology that integrates the Beidou satellite navigation system with an inertial measurement unit, automatically switching navigation modes based on signal quality. A dual-link redundant communication system is also constructed: the primary link utilizes a 5G private network to connect to mobile edge computing nodes, while the backup link utilizes a UWB ultra-wideband communication module. S06. Initial modeling of pollutant concentration fields: Processing discrete monitoring point data based on the inverse distance weighted algorithm: Determine the set of monitoring points involved in interpolation, calculate the Euclidean distance between the target point and each monitoring point, and generate an initial two-dimensional concentration field by weighted summation of the concentration values ​​at each monitoring point. S07. Concentration Field Optimization and Visualization: Optimize the initial concentration field using the Kriging algorithm. Analyze spatial variation characteristics using the variogram, determine semivariogram model parameters, calculate target point predictions based on the unbiased optimal estimation principle, and generate a three-dimensional AQI heat map. S08. Adaptive Route Planning: Identifies high-pollution areas based on a 3D AQI heat map and automatically plans complex routes: Spiral routes are used for refined monitoring in pollution hotspots, while grid routes are used to improve coverage efficiency in non-hotspot areas. S09. Pollutant Diffusion Prediction: After receiving regional monitoring data, the pollutant diffusion rate is predicted using a random forest algorithm based on historical pollution diffusion data and real-time meteorological conditions.

2. The environmental monitoring method based on a low-altitude aircraft according to claim 1, characterized in that: The sensor cabin hardware architecture in S02 includes: S21. Install the temperature and humidity sensor, air pressure sensor, and ultrasonic anemometer on three sides of the cabin. Secure the sensors with aluminum alloy brackets, which are connected to the cabin frame with screws. S22. A micro differential absorption spectrometer and PM2.5 laser sensor are mounted on the same side. The spectrometer is bonded to the cabin shell via a thermally conductive silicone pad. A guide groove is designed around the laser sensor. An active air intake system is constructed using a micro air pump, and a waterproof and breathable membrane is installed at the air intake. S23. The lidar and multispectral camera are installed on the cabin roof, using an electronic level to assist with installation. All sensor cables are secured through cable management troughs inside the cabin, using shielded twisted-pair cables for signal transmission, and shielded metal connectors at the interfaces. S24. Sandblast the surface of the cabin and apply a super-hydrophobic coating using chemical vapor deposition technology. S25. Each sensor is mounted using four shock-absorbing pads arranged in a rectangular pattern.

3. The environmental monitoring method based on a low-altitude aircraft according to claim 1, characterized in that: The dynamic real-time calibration of S03 is implemented by an edge computing processing unit, including: S31. Build a BP neural network model with three hidden layers. The first hidden layer has 64 neurons, the second hidden layer has 32 neurons, and the third hidden layer has 16 neurons. The input layer receives the original temperature and humidity signals and the temperature compensation coefficient, and the output layer outputs the calibration parameters. S32. Generate a standard environment using a high-precision temperature and humidity calibration chamber, and simultaneously collect the sensor's raw output and standard values ​​to construct a training dataset. S33. The Adam optimization algorithm is used to train the neural network for 500 cycles. The edge computing unit uses a 50ms cycle to input real-time data into the trained model and output calibration parameters.

4. The environmental monitoring method based on a low-altitude aircraft according to claim 1, characterized in that: The specific steps of generating the three-dimensional environment grid model in S04 include: S41. Using the extended Kalman filter algorithm, the current state is predicted based on the state estimate at the previous moment in the prediction phase; S42. In the update phase, the actual measured value is integrated with the predicted value; S43. Repeat the iterative calculations of S41-S42 to make the timestamp alignment error ≤ ±10 μs; S44. Fuse the spatiotemporally aligned lidar point cloud with camera image data and extract obstacle contours using a point cloud segmentation algorithm. S45. Use the optical flow method to calculate the flow field vector, with a 10cm 3 The grid resolution generates a three-dimensional environment grid model including obstacle distribution and flow field characteristics.

5. The environmental monitoring method based on a low-altitude aircraft according to claim 1, characterized in that: The implementation of the navigation communication system in S05 includes: S51. Use a combined navigation solution with a BeiDou-3 RTK receiver and a high-precision IMU. S52. Receive differential correction data from ground-based augmentation stations to achieve centimeter-level positioning with a data update rate of 10Hz; S53. Automatically switch to visual SLAM navigation mode when Beidou signal quality degrades; S54 uses the ORB-SLAM3 algorithm framework, equipped with an 8K camera, and extracts feature points at 500fps using FAST corner detection and BRIEF descriptors. S55. Combine IMU data with visual-inertial tight coupling to achieve monocular visual positioning accuracy of ±0.5m, and verify navigation data consistency through Kalman filtering; S56. The primary link uses a module that supports the 5G NR standard to connect to the mobile edge computing node, configured with a 20MHz bandwidth carrier, and an end-to-end latency of less than 10ms; S57. The backup link uses a UWB module (operating frequency band 3.1-10.6 GHz) with DS-UWB modulation technology, automatically switching if the primary link is interrupted. S58. Integrates a 2-18GHz broadband spectrum detection module that scans the electromagnetic environment with a 10ms cycle and triggers frequency band switching when the interference signal power is greater than -60dBm.

Images