An air route weather risk intelligent evaluation and adaptive avoidance method and system

Abstract

This application provides a method and system for intelligent assessment and adaptive avoidance of airway meteorological risks. The method includes: acquiring multi-source meteorological data and constructing a three-dimensional meteorological field of the target airspace based on the multi-source meteorological data; establishing an aircraft type threshold model reflecting the aircraft's tolerance to different meteorological elements based on the aircraft's airworthiness information; generating a comprehensive risk field characterizing the distribution of meteorological risks within the target airspace based on the three-dimensional meteorological field and the aircraft type threshold model; extrapolating the spatiotemporal evolution trend of the comprehensive risk field using a risk spatiotemporal propagation model with embedded physical constraints to obtain a predicted risk field; acquiring the aircraft's planned flight path and performing path integration of the predicted risk field along the planned flight path to obtain a cumulative risk value for the flight path; comparing the cumulative risk value with a dynamic safety threshold and generating avoidance commands for the flight control system based on the comparison result.

Description

Technical Field

[0001] This invention relates to the field of aircraft route safety technology, and in particular to a method and system for intelligent assessment and adaptive avoidance of route meteorological risks. Background Technology

[0002] In recent years, with the rapid development of the low-altitude economy, drones have been widely used in urban delivery, industrial inspection, logistics transportation, emergency rescue, and other scenarios. The meteorological environment in low-altitude airspace is complex and changeable, with risks such as wind shear, turbulence, strong crosswinds, heavy precipitation, and even micro-scale meteorological changes. These factors pose a serious threat to the flight safety of drones.

[0003] Currently, the main solutions for addressing weather risks along drone flight paths include the following: 1. Manual avoidance based on weather forecasts: This method relies on pilots or dispatchers making subjective judgments based on weather forecast information, which is not timely and is highly subjective.

[0004] 2. Static threshold judgment: Setting a single meteorological element threshold (such as prohibiting flight when the wind speed exceeds 8m / s) is simple but lacks flexibility, cannot cope with the risks of complex weather combinations, and does not take into account the differences in wind resistance capabilities of different aircraft models.

[0005] 3. Weighted meteorological cost based on path planning: The path planning algorithm treats meteorological information as a simple weighted cost item, which fails to provide a detailed model of the three-dimensional spatial distribution and temporal evolution of meteorological risk.

[0006] 4. Application of general weather forecasting models: These models are usually used for macro-meteorological analysis, but their spatial and temporal resolutions are insufficient to meet the needs of low-altitude UAV flight path safety analysis.

[0007] Therefore, the aforementioned existing technology has the following drawbacks: 1. Weather forecast results are not coupled with the airworthiness of the UAV itself, making it impossible to assess the true risk of a specific UAV model under specific weather conditions; 2. Lack of detailed modeling capabilities for the three-dimensional micro-meteorological environment in low-altitude airspace (especially the 0-300 meter altitude layer); 3. It is impossible to make dynamic and quantitative propagation inferences about the spatiotemporal evolution trend of meteorological risks in the short term; 4. The meteorological risk assessment results were not integrated with the specific flight route for analysis, making it difficult to quantify the cumulative risk of the entire route; 5. The output results are mostly risk levels or alarm information, which cannot be directly converted into flight control parameters that the flight control system can execute. Summary of the Invention

[0008] The purpose of this invention is to provide a method and system for intelligent assessment and adaptive avoidance of airway meteorological risks, aiming to solve the above-mentioned problems in the prior art.

[0009] This invention provides a method for intelligent assessment and adaptive avoidance of airway meteorological risks, including: Acquire multi-source meteorological data and construct a three-dimensional meteorological field for the target airspace based on the multi-source meteorological data; Establish a threshold model based on the aircraft's airworthiness information to reflect the aircraft's tolerance to different meteorological elements; A comprehensive risk field characterizing the distribution of meteorological risks within the target airspace is generated based on the three-dimensional meteorological field and the aircraft type threshold model. The spatiotemporal evolution trend of the comprehensive risk field is deduced using a risk spatiotemporal propagation model with embedded physical constraints, and the predicted risk field is obtained. The planned flight path of the aircraft is obtained, and the predicted risk field is integrated along the planned flight path to obtain the cumulative risk value of the flight path. The cumulative risk value of the route is compared with the dynamic safety threshold, and avoidance instructions for the flight control system are generated based on the comparison result.

[0010] This invention provides an intelligent assessment and adaptive avoidance system for airway meteorological risks, comprising: The meteorological field construction module is used to acquire multi-source meteorological data and construct a three-dimensional meteorological field of the target airspace based on the multi-source meteorological data. The aircraft type threshold configuration module is used to establish an aircraft type threshold model that reflects the aircraft's tolerance to different meteorological elements based on the aircraft's airworthiness information. The risk assessment module is used to generate a comprehensive risk field characterizing the distribution of meteorological risks within the target airspace based on the three-dimensional meteorological field and the aircraft type threshold model. The risk extrapolation module is used to extrapolate the spatiotemporal evolution trend of the comprehensive risk field using a risk spatiotemporal propagation model with embedded physical constraints, and obtain the predicted risk field. The route analysis module is used to obtain the planned route of the aircraft, and to perform path integration of the predicted risk field along the planned route to obtain the route cumulative risk value. The decision output module is used to compare the cumulative risk value of the route with the dynamic safety threshold, and generate avoidance instructions for the flight control system based on the comparison result.

[0011] This invention also provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the above-described intelligent assessment and adaptive avoidance method for airway meteorological risks.

[0012] This invention also provides a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor, implements the steps of the above-described intelligent assessment and adaptive avoidance method for airway meteorological risks.

[0013] The embodiments of this invention can include the following beneficial effects: First, a three-dimensional meteorological risk field is constructed, which integrates multi-source meteorological data and UAV airworthiness threshold matrix, and can finely characterize the meteorological risk distribution in low-altitude airspace. On this basis, a spatiotemporal propagation inference model based on physical equations (such as advection-diffusion equations) or physical-guided neural networks is introduced to dynamically predict the meteorological risk field in the future. Finally, by performing integral coupling analysis between the planned flight path and the predicted risk field, the comprehensive risk of the entire flight path is quantified, and control parameters that can be directly used in the flight control system are generated, thereby realizing an intelligent risk inference closed loop from meteorological data to flight control commands, so as to improve the reliability and real-time performance of UAV flight path safety decisions. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in one or more embodiments of this specification or in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart of the intelligent assessment and adaptive avoidance method for airway meteorological risks according to an embodiment of the present invention; Figure 2 This is a timing diagram of an embodiment of the present invention; Figure 3 This is a schematic diagram of the intelligent assessment and adaptive avoidance system for airway meteorological risks according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the overall system structure according to an embodiment of the present invention. Detailed Implementation

[0016] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0017] Method Implementation Examples According to embodiments of the present invention, a method for intelligent assessment and adaptive avoidance of airway meteorological risks is provided. Figure 1 This is a flowchart of the intelligent assessment and adaptive avoidance method for airway meteorological risks according to an embodiment of the present invention, as follows: Figure 1 As shown, the intelligent assessment and adaptive avoidance method for airway meteorological risks according to an embodiment of the present invention specifically includes: Step S101 involves acquiring multi-source meteorological data and constructing a three-dimensional meteorological field for the target airspace based on the multi-source meteorological data; specifically including: Collect multi-source meteorological data; wherein, the multi-source meteorological data includes radar meteorological data, ground meteorological station data, and meteorological data collected by airborne sensors of aircraft; The multi-source meteorological data were spatiotemporally aligned and quality controlled, and multi-source fusion was performed using a Kalman filter algorithm. Acquire geographic information data corresponding to the target airspace; wherein, the geographic information data includes at least one of digital surface model, building white model and surface roughness; The fused meteorological data and the geographic information data are spatially aligned and stitched together to obtain the input tensor; The input tensor is fed into a pre-trained 3D super-resolution reconstruction network to output a 3D meteorological field of the target spatial domain; wherein, the 3D meteorological field includes the spatial and temporal distributions of wind speed vector, wind direction, turbulence intensity, precipitation intensity, and temperature parameters.

[0018] Step S102: Establish a model based on the aircraft's airworthiness information to reflect the aircraft's tolerance to different meteorological elements; specifically including: Obtain the aircraft type identifier of the aircraft currently performing a mission; The corresponding airworthiness configuration file is read from the pre-stored aircraft capability database based on the aircraft type identifier; The airworthiness configuration file is parsed to extract the tolerance boundary parameters of the aircraft in various meteorological parameter dimensions. A model threshold matrix is ​​constructed based on the tolerance boundary parameters, and the model threshold matrix is ​​used as the model threshold model; The threshold matrix for the aircraft model includes the maximum tolerable crosswind component, the maximum tolerable wind shear intensity, the maximum tolerable turbulence level, and the maximum tolerable precipitation intensity.

[0019] Step S103: Generate a comprehensive risk field characterizing the meteorological risk distribution within the target airspace based on the three-dimensional meteorological field and the aircraft type threshold model; specifically including: Traverse each spatial grid point within the target airspace and obtain the crosswind speed, wind shear intensity, turbulence intensity, and precipitation intensity of the current grid point in the three-dimensional meteorological field as meteorological elements to be evaluated; Obtain the tolerance boundary parameters corresponding to the meteorological element to be evaluated in the threshold model of the aircraft model; Calculate the normalized risk component of each meteorological element to be assessed relative to the corresponding tolerance boundary parameter; The normalized risk components are weighted and merged according to preset weights to obtain the comprehensive risk value of the current grid point. Traverse all grid points to generate a comprehensive risk field composed of the combined risk values ​​of each grid point.

[0020] Step S104 involves using a risk spatiotemporal propagation model with embedded physical constraints to deduce the spatiotemporal evolution trend of the comprehensive risk field, thereby obtaining the predicted risk field; specifically including: The current comprehensive risk field and the wind speed vector in the three-dimensional meteorological field are used as input conditions, and the spatiotemporal coordinates of the target prediction time are used as query points. These are input into the pre-trained risk spatiotemporal propagation model, and the predicted risk value corresponding to the query point is output. The risk spatiotemporal propagation model is a physically guided neural network, and its loss function incorporates the advection-diffusion equation of the risk value as a physical constraint. Traverse all grid points within the target airspace to generate a predicted risk field composed of the predicted risk values ​​of each grid point.

[0021] Step S105: Obtain the planned flight path of the aircraft, and perform path integration of the predicted risk field along the planned flight path to obtain the cumulative risk value of the flight path; specifically including: Obtain the planned route of the aircraft; wherein the planned route consists of a series of waypoints, each waypoint including its position coordinates and estimated arrival time; Spatially match the coordinates of each waypoint on the planned route with the predicted risk field, and extract the predicted risk value corresponding to each waypoint. Based on the predicted risk value of each waypoint and the distance between adjacent waypoints, the path integral is performed along the planned route to obtain the cumulative risk value of the route.

[0022] Step S106 involves comparing the cumulative risk value of the flight path with the dynamic safety threshold, and generating avoidance commands for the flight control system based on the comparison result. Specifically, this includes: Obtain a dynamic security threshold; wherein the dynamic security threshold is adaptively adjusted based on task context information; Compare the cumulative risk value of the route with the dynamic safety threshold; If the cumulative risk value of the route is less than the dynamic safety threshold, the route is determined to be safe, and a continue flight command is output. If the cumulative risk value of the route is greater than or equal to the dynamic safety threshold, the route is determined to have unacceptable risks. An avoidance instruction is generated based on the spatial distribution characteristics of the predicted risk field and sent to the flight control system for execution. The avoidance instruction includes one or more of the following: route fine-tuning instruction, altitude adjustment instruction, speed limit instruction, route replanning instruction, or return-to-origin instruction.

[0023] The following details the intelligent assessment and adaptive avoidance method for airway meteorological risks according to embodiments of the present invention, such as... Figure 2 As shown, the above technical solutions of the embodiments of the present invention will be described in detail.

[0024] This invention proposes an intelligent meteorological risk reasoning method for UAV route safety, comprising the following steps: 1. Multi-source meteorological data acquisition and fusion: Real-time acquisition of radar meteorological data, ground meteorological station data, UAV airborne micro-meteorological sensor data, etc.

[0025] 2. Three-dimensional meteorological field modeling: Based on the collected multi-source data, a continuous three-dimensional meteorological field is constructed using a deep learning-based super-resolution reconstruction network. Where x, y, z are spatial coordinates and t is time; the meteorological field includes multiple meteorological parameters such as wind speed, wind direction, turbulence intensity, precipitation intensity, and temperature.

[0026] 3. Airworthiness Threshold Matrix Construction: Construct an airworthiness threshold matrix for different UAV models. This matrix should include key parameters such as the maximum tolerable crosswind component, maximum wind shear intensity, maximum turbulence level, and maximum precipitation intensity. A code example is shown below: T_model_droneA= { CrossWind_max=8m / s Shear_max=5m / s Turbulence_max=4 Rain_max=6mm / h } 4. Construction of meteorological risk function: based on three-dimensional meteorological field and model threshold matrix Construct a comprehensive meteorological risk function This function integrates the degree of exceedance of various meteorological elements relative to the aircraft model threshold through weighted normalization, and outputs a dimensionless risk value (for example, normalized to the interval [0, 1], where >1 indicates unacceptable risk).

[0027] 5. Spatiotemporal propagation reasoning of meteorological risks: A Physics-Informed Neural Network (PINN) is introduced as a spatiotemporal propagation model for risks, to analyze the current risk field. The network infers its future evolution. It directly embeds the advection-diffusion equation of the risk field as a physical constraint into the loss function and uses a small amount of real-time observation data for semi-supervised learning training.

[0028] Compared to purely data-driven models, PINN can more accurately learn the risk propagation patterns under the influence of complex terrain and wind fields, maintaining good prediction accuracy even in areas with sparse meteorological data. The network ultimately outputs a predicted future. Risk field at any moment .

[0029] 6. Route Risk Coupling Analysis: Obtain the planned flight path of the UAV. Path integration is performed along the flight path to predict the risk field: ; Compare the calculation results with the dynamic adaptive safety threshold. Comparison, among which It is a function of the mission context c (including dynamic factors such as mission urgency, remaining drone battery power, communication link quality, and battery status). For example, the safety threshold for an emergency medical transport mission can be dynamically increased by 30% to strike a balance between risk and mission necessity.

[0030] 7. Flight control command generation and output: If Exceeding the dynamic security threshold Based on the spatial distribution characteristics of the risk, an avoidance strategy is generated, including but not limited to route replanning weight parameters, altitude adjustment commands, speed limit commands, or return-to-home commands, and sent to the UAV flight control system for execution through the flight control interface module.

[0031] This embodiment uses an urban drone delivery scenario as an example to further illustrate the present invention.

[0032] Step 1: Multi-source meteorological data acquisition and fusion The system acquires the following multi-source data in real time through the meteorological data acquisition module: (1) Radar meteorological data: provides information on precipitation intensity and regional wind field over a wide area, with a time resolution of 5-10 minutes and a spatial resolution of about 1 kilometer; (2) Ground meteorological station data: Provides point observation data with high temporal resolution (1 minute), including wind speed, wind direction, temperature, precipitation, etc.; (3) Airborne micro-meteorological data of UAVs: transmitted back by UAVs that are currently performing or have performed missions in the past, providing real meteorological samples along the flight route.

[0033] After undergoing spatiotemporal alignment and quality control, the aforementioned data was fused from multiple sources using a Kalman filter algorithm, providing a unified input for subsequent 3D meteorological field modeling.

[0034] Step 2: 3D meteorological field modeling Based on collected multi-source data, a continuous three-dimensional meteorological field is constructed using a deep learning-based super-resolution reconstruction network. The details are as follows: (1) Input data preparation The collected multi-source heterogeneous data was spatiotemporally aligned and preprocessed. Radar meteorological data: provides information on precipitation intensity and wind field over a wide area; Ground weather station data: provides high temporal resolution point observation data (wind speed, wind direction, temperature, precipitation, etc.); UAV-borne micro-meteorological data: provides real-time meteorological sampling data along the flight path; High-resolution geographic information data, including digital surface models (DSM), building white models, surface roughness, etc., are stored offline as prior structural information.

[0035] (2) Super-resolution reconstructed network architecture A 3D super-resolution generative adversarial network (3D SRGAN) is used as the core reconstruction model, and its structure includes: The generator employs a 3D residual dense connection network (3D RRDB). The input is a spliced ​​tensor of a low-resolution meteorological field (from numerical weather prediction or radar assimilation products) and a high-resolution geographic feature map. It extracts depth features through multiple residual dense modules and then uses 3D sub-pixel convolutional layers to gradually improve the spatial resolution by 4-8 times, finally outputting a high-resolution 3D meteorological field.

[0036] Discriminator: A three-dimensional PatchGAN structure is used to distinguish between true and false data in the generated high-resolution field block by block, which forces the generator to produce more realistic and sharper micro-meteorological details (such as the turbulent structure of the wake region of a building).

[0037] (3) Embedding of physical constraints Atmospheric physics equations are embedded as soft constraints into the network training process to enhance the physical consistency of the output field. A. Mass conservation constraint: For the incompressible flow approximation, wind field divergence... ; B. Turbulent dissipation constraint: The turbulence intensity and the wind speed gradient satisfy an empirical relationship; C. Thermodynamic constraints: The rate of temperature decrease with altitude conforms to atmospheric stability conditions.

[0038] These constraints are added to the total loss in the form of an additional loss function, enabling the network to follow physical laws while fitting the data.

[0039] (4) Training strategies Training data generation: Computational fluid dynamics (CFD) simulations (such as large eddy simulation) are used to generate high-resolution 3D meteorological fields for typical urban scenes as real labels, while downsampling and sparsification are applied to simulate low-resolution inputs.

[0040] Two-stage training: First, pre-training is performed using pure simulation data to learn general meteorological super-resolution mapping relationships; then, fine-tuning is performed using a small amount of real observation data (such as UAV airborne data) to adapt to actual atmospheric conditions.

[0041] (5) Online reasoning When the system is running in real time, the following steps are performed: Spatially align the real-time acquired low-resolution weather forecast field with the offline stored high-resolution geographic feature map; If real-time airborne observation data exists, it is fused and corrected with the low-resolution field through Kalman filtering or weighted averaging. The fused input tensor is fed into the pre-trained generator network to perform one forward propagation. The generator outputs a high-resolution three-dimensional meteorological field. Covering the target airspace range; Post-process the output field to ensure that all meteorological parameters are within a reasonable range (e.g., wind speed is non-negative and turbulence intensity is between 0 and 1). The completed three-dimensional meteorological field is then transferred to the downstream risk function construction module.

[0042] (6) Output results The final output three-dimensional meteorological field It is a five-dimensional tensor (three-dimensional space + one-dimensional time + one-dimensional meteorological parameter channel), at each grid point The above includes the following meteorological parameters: wind speed vector Wind direction, turbulence intensity, precipitation intensity, and temperature.

[0043] Step 3: Constructing the Airworthiness Threshold Matrix for Aircraft Type The system reads the airworthiness configuration file of the drone model currently performing the mission and constructs a threshold matrix. This matrix defines safety boundaries for various meteorological parameters for different aircraft models: ; in, To withstand the maximum crosswind load, To withstand the maximum wind shear strength, The maximum tolerable turbulence level, ranging from [0, 1]; The maximum tolerable rainfall intensity.

[0044] For example, for a certain model of delivery drone A, its threshold matrix is ​​as follows: maximum crosswind: 8 m / s; maximum wind shear: 5 m / s per 100 m; maximum turbulence: 0.4; maximum precipitation: 6 mm / h. The threshold matrices for different models (such as inspection drones, logistics drones, and emergency rescue drones) can be obtained through wind tunnel testing, flight testing, or model manuals and are pre-stored in the system database.

[0045] Step 4: Constructing the meteorological risk function The three-dimensional meteorological field constructed in step 2 and the model threshold matrix in step 3 Construct a comprehensive meteorological risk function .

[0046] (1) Single-factor risk calculation For each grid point, first calculate the normalized risk component of each meteorological element relative to the aircraft type threshold: Crosswind risk: ; Wind shear risk: ; Turbulence risk: ; Precipitation risk: ; (2) Multi-element integration The risk components are combined using a weighted sum method: ; Among them, weight parameters , , , satisfy It can be dynamically adjusted according to the characteristics of the aircraft or the flight phase (such as increasing the crosswind weight during takeoff and landing).

[0047] (3) Risk level determination ≤0.6: Low risk, flight is permitted; 0.6< ≤0.9: Medium risk, worth noting; 0.9< ≤1.0: High risk, requires careful assessment; >1.0: Risk exceeding the threshold, entry prohibited.

[0048] Step 5: Spatiotemporal propagation reasoning of meteorological risks Introducing a physically guided neural network as a risk spatiotemporal propagation model to analyze the current risk field. We will infer its future evolution.

[0049] (1) Physical equations The spatiotemporal evolution of the risk field follows the advection-diffusion equation: ; In the formula, The wind field vector is derived from the three-dimensional meteorological field in step 2; The risk gradient represents the spatial direction of risk variation. For the risk Laplace operator, it represents the diffusion effect of risk; The advection coefficient controls the intensity of wind field transport of risk. The diffusion coefficient represents the ability to control the self-diffusion of risk.

[0050] (2) PINN network architecture The above partial differential equations are solved using a physics-guided neural network: Input layer: Spatial coordinates and time t; Hidden layers: multi-layer fully connected network, using a sinusoidal activation function to better fit high-frequency features; Output layer: Predicted risk value.

[0051] (3) Loss function The loss function consists of two parts: A. Data loss: the mean square error between the predicted value and the current risk field observation; B. Physical loss: the penalty term for the predicted value deviating from the advection-diffusion equation residual.

[0052] ; in These are the physical constraint weighting coefficients.

[0053] (4) Reasoning process A. Based on the current time The risk field and wind field are used as initial conditions; B. Predict the target time The spatiotemporal coordinates are input into PINN; C. Network forward propagation, outputting predicted risk values; D. Traverse all grid points to generate the future risk field. ; E. Normalize the prediction results to ensure... . (5) Result output Output the future Three-dimensional risk field at a moment (5 - 15 minutes) to describe the spatio-temporal evolution trend of meteorological risks.

[0054] Step 6: Route risk coupling analysis (1) Route parameterization Obtain the planned route path of the UAV , where s is the arc length parameter. The route consists of a series of waypoints, and each waypoint contains position coordinates and the estimated arrival time t. (2) Path risk integration Perform path integration on the predicted risk field along the route: ; where, is the total length of the route, is the estimated time to reach position s.

[0055] Discretized calculation: ; where N is the number of waypoints, is the distance between adjacent waypoints.

[0056] (3) Safety threshold comparison Compare the calculated with the preset safety threshold Th: If < Th, determine that the route is safe and output a continue flight instruction; if ≥ Th, determine that there is an unacceptable risk on the route and trigger a risk response strategy.

[0057] (4) Dynamic threshold adjustment The safety threshold Th can be dynamically adjusted according to the task context: ; where, is the reference threshold (take 0.8), is the adjustment function, is the context factor, including: task urgency (e.g., a 20% increase for medical transportation tasks); remaining battery power of the UAV (lower the threshold when the power is insufficient); communication link quality (lower the threshold when the link is poor); task period (lower the threshold for night flights).

[0058] Step 7: Flight control instruction generation and output When it is determined in Step 6 that the route risk exceeds the threshold, the control instruction output module generates an avoidance strategy according to the spatial distribution characteristics of the risk.

[0059] (1) Risk feature analysis Analyze 's spatial distribution and identify the following features: A. Risk Center: The location with the highest risk value; B. Risk gradient: The direction in which risk decreases the fastest; C. Risk boundary: The boundary of the acceptable risk area.

[0060] (2) Generation of avoidance strategies Based on risk characteristics, the following avoidance instructions are generated in priority, as shown in Table 1: Table 1 Avoidance Strategies

[0061] (3) Command encapsulation and transmission The generated flight control commands are encapsulated according to standard flight control interface protocols (such as MAVLink) and include: command type (such as SET_POSITION, SET_ALTITUDE, SET_VELOCITY); target parameters (such as target altitude, target speed, target waypoint coordinates); execution priority and timestamp.

[0062] The encapsulated instructions are sent to the UAV flight control system for execution via a communication link.

[0063] (4) Continuous monitoring and closed-loop feedback After sending the evasion command, the system continuously monitors the drone's execution status and changes in weather risk: if the drone successfully evades the evasion and the risk of the new route drops below the threshold, normal monitoring is restored; if the risk continues to worsen or the drone deviates from the expected path, a higher-level response (such as emergency return) is triggered.

[0064] Preferably, the risk function in this embodiment of the invention may also employ a fuzzy logic reasoning model.

[0065] In summary, the embodiments of the present invention realize a complete intelligent risk reasoning closed loop from multi-source meteorological data to flight control commands. The system can accurately depict the three-dimensional meteorological field at low altitude, quantify the adaptability of aircraft models, predict the spatiotemporal evolution of risks, quantify the cumulative risks along the flight path, and generate executable flight control commands, thereby significantly improving the flight safety decision-making ability of UAVs under complex meteorological conditions.

[0066] System Implementation Examples According to embodiments of the present invention, an intelligent assessment and adaptive avoidance system for airway meteorological risks is provided. Figure 3 This is a schematic diagram of the intelligent assessment and adaptive avoidance system for airway meteorological risks according to an embodiment of the present invention, as shown below. Figure 3 As shown, the intelligent assessment and adaptive avoidance system for airway meteorological risks according to an embodiment of the present invention specifically includes: The meteorological field construction module 30 is used to acquire multi-source meteorological data and construct a three-dimensional meteorological field of the target airspace based on the multi-source meteorological data. The aircraft type threshold configuration module 31 is used to establish an aircraft type threshold model that reflects the aircraft's tolerance to different meteorological elements based on the aircraft's airworthiness information. Risk assessment module 32 is used to generate a comprehensive risk field characterizing the distribution of meteorological risks within the target airspace based on the three-dimensional meteorological field and the aircraft type threshold model; The risk extrapolation module 33 is used to extrapolate the spatiotemporal evolution trend of the comprehensive risk field using a risk spatiotemporal propagation model with embedded physical constraints, and obtain the predicted risk field. The route analysis module 34 is used to obtain the planned route of the aircraft and perform path integration of the predicted risk field along the planned route to obtain the route cumulative risk value. The decision output module 35 is used to compare the cumulative risk value of the route with the dynamic safety threshold, and generate avoidance instructions for the flight control system based on the comparison result.

[0067] Furthermore, the system proposed in the embodiments of the present invention is specifically as follows: Figure 4 As shown, it includes: a meteorological data acquisition module, a 3D meteorological field construction module, an airworthiness risk function construction module, a meteorological risk spatiotemporal propagation inference module, a route risk coupling analysis module, and a control command output module. Its data flow is: meteorological data acquisition → 3D meteorological modeling → risk function construction → spatiotemporal inference → route coupling analysis → control output.

[0068] The embodiments of the present invention are system embodiments corresponding to the above method embodiments. The specific operation of each module can be understood by referring to the description of the method embodiments, and will not be repeated here.

[0069] In summary, compared with the prior art, the beneficial effects of the embodiments of the present invention include: 1. Transform meteorological data into flight route safety risk reasoning results; 2. Supports airworthiness risk modeling based on aircraft model differences; 3. Achieve dynamic spatiotemporal forecasting of meteorological risks; 4. Risk scores can be quantified for the entire route; 5. Output executable control parameters; 6. Improve the emergency meteorological response capabilities of drones.

[0070] Device Example 1 This invention provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it performs the steps described in the method embodiment.

[0071] Device Example 2 This invention provides a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor, performs the steps described in the method embodiment.

[0072] The computer-readable storage media described in this embodiment include, but are not limited to, ROM, RAM, disk, or optical disk.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent assessment and adaptive avoidance of weather risk for a route, characterized in that, include: Acquire multi-source meteorological data and construct a three-dimensional meteorological field for the target airspace based on the multi-source meteorological data; Establish a threshold model based on the aircraft's airworthiness information to reflect the aircraft's tolerance to different meteorological elements; A comprehensive risk field characterizing the distribution of meteorological risks within the target airspace is generated based on the three-dimensional meteorological field and the aircraft type threshold model. The spatiotemporal evolution trend of the comprehensive risk field is deduced using a risk spatiotemporal propagation model with embedded physical constraints, and the predicted risk field is obtained. The planned flight path of the aircraft is obtained, and the predicted risk field is integrated along the planned flight path to obtain the cumulative risk value of the flight path. The cumulative risk value of the route is compared with the dynamic safety threshold, and avoidance instructions for the flight control system are generated based on the comparison result.

2. The method according to claim 1, characterized in that, Acquiring multi-source meteorological data and constructing a three-dimensional meteorological field for the target airspace based on the multi-source meteorological data specifically includes: Collect multi-source meteorological data; wherein, the multi-source meteorological data includes radar meteorological data, ground meteorological station data, and meteorological data collected by airborne sensors of aircraft; The multi-source meteorological data were spatiotemporally aligned and quality controlled, and multi-source fusion was performed using a Kalman filter algorithm. Acquire geographic information data corresponding to the target airspace; wherein, the geographic information data includes at least one of digital surface model, building white model and surface roughness; The fused meteorological data and the geographic information data are spatially aligned and stitched together to obtain the input tensor; The input tensor is fed into a pre-trained 3D super-resolution reconstruction network to output a 3D meteorological field of the target spatial domain; wherein, the 3D meteorological field includes the spatial and temporal distributions of wind speed vector, wind direction, turbulence intensity, precipitation intensity, and temperature parameters.

3. The method according to claim 1, characterized in that, The establishment of an aircraft type threshold model based on aircraft type airworthiness information, reflecting the aircraft's tolerance to different meteorological elements, specifically includes: Obtain the aircraft type identifier of the aircraft currently performing a mission; The corresponding airworthiness configuration file is read from the pre-stored aircraft capability database based on the aircraft type identifier; The airworthiness configuration file is parsed to extract the tolerance boundary parameters of the aircraft in various meteorological parameter dimensions. A model threshold matrix is ​​constructed based on the tolerance boundary parameters, and the model threshold matrix is ​​used as the model threshold model; The threshold matrix for the aircraft model includes the maximum tolerable crosswind component, the maximum tolerable wind shear intensity, the maximum tolerable turbulence level, and the maximum tolerable precipitation intensity.

4. The method according to claim 1, characterized in that, The comprehensive risk field representing the distribution of meteorological risks within the target airspace, generated based on the three-dimensional meteorological field and the aircraft type threshold model, specifically includes: Traverse each spatial grid point within the target airspace and obtain the crosswind speed, wind shear intensity, turbulence intensity, and precipitation intensity of the current grid point in the three-dimensional meteorological field as meteorological elements to be evaluated; Obtain the tolerance boundary parameters corresponding to the meteorological element to be evaluated in the threshold model of the aircraft model; Calculate the normalized risk component of each meteorological element to be assessed relative to the corresponding tolerance boundary parameter; The normalized risk components are weighted and merged according to preset weights to obtain the comprehensive risk value of the current grid point. Traverse all grid points to generate a comprehensive risk field composed of the combined risk values ​​of each grid point.

5. The method according to claim 1, characterized in that, The spatiotemporal evolution trend of the comprehensive risk field is deduced using a risk spatiotemporal propagation model with embedded physical constraints, resulting in a predicted risk field that specifically includes: The current comprehensive risk field and the wind speed vector in the three-dimensional meteorological field are used as input conditions, and the spatiotemporal coordinates of the target prediction time are used as query points. These are input into the pre-trained risk spatiotemporal propagation model, and the predicted risk value corresponding to the query point is output. The risk spatiotemporal propagation model is a physically guided neural network, and its loss function incorporates the advection-diffusion equation of the risk value as a physical constraint. Traverse all grid points within the target airspace to generate a predicted risk field composed of the predicted risk values ​​of each grid point.

6. The method according to claim 1, characterized in that, Obtaining the planned flight path of the aircraft and integrating the predicted risk field along the planned flight path to obtain the cumulative risk value of the flight path specifically includes: Obtain the planned route of the aircraft; wherein the planned route consists of a series of waypoints, each waypoint including its position coordinates and estimated arrival time; Spatially match the coordinates of each waypoint on the planned route with the predicted risk field, and extract the predicted risk value corresponding to each waypoint. Based on the predicted risk value of each waypoint and the distance between adjacent waypoints, the path integral is performed along the planned route to obtain the cumulative risk value of the route.

7. The method according to claim 1, characterized in that, The process of comparing the cumulative risk value of the flight path with a dynamic safety threshold and generating avoidance commands for the flight control system based on the comparison result specifically includes: Obtain a dynamic security threshold; wherein the dynamic security threshold is adaptively adjusted based on task context information; Compare the cumulative risk value of the route with the dynamic safety threshold; If the cumulative risk value of the route is less than the dynamic safety threshold, the route is determined to be safe, and a continue flight command is output. If the cumulative risk value of the route is greater than or equal to the dynamic safety threshold, the route is determined to have unacceptable risks. An avoidance instruction is generated based on the spatial distribution characteristics of the predicted risk field and sent to the flight control system for execution. The avoidance instruction includes one or more of the following: route fine-tuning instruction, altitude adjustment instruction, speed limit instruction, route replanning instruction, or return-to-origin instruction.

8. A smart assessment and adaptive avoidance system for airway meteorological risks, characterized in that, include: The meteorological field construction module is used to acquire multi-source meteorological data and construct a three-dimensional meteorological field of the target airspace based on the multi-source meteorological data. The aircraft type threshold configuration module is used to establish an aircraft type threshold model that reflects the aircraft's tolerance to different meteorological elements based on the aircraft's airworthiness information. The risk assessment module is used to generate a comprehensive risk field characterizing the distribution of meteorological risks within the target airspace based on the three-dimensional meteorological field and the aircraft type threshold model. The risk extrapolation module is used to extrapolate the spatiotemporal evolution trend of the comprehensive risk field using a risk spatiotemporal propagation model with embedded physical constraints, and obtain the predicted risk field. The route analysis module is used to obtain the planned route of the aircraft, and to perform path integration of the predicted risk field along the planned route to obtain the route cumulative risk value. The decision output module is used to compare the cumulative risk value of the route with the dynamic safety threshold, and generate avoidance instructions for the flight control system based on the comparison result.

9. An electronic device, characterized in that, include: The memory, the processor, and the computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the route weather risk intelligent assessment and adaptive avoidance method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an information transmission implementation program, which, when executed by a processor, implements the steps of the route meteorological risk intelligent assessment and adaptive avoidance method as described in any one of claims 1-7.

Images