Real-time weather risk assessment method for low-altitude flight safety

Abstract

The application discloses a real-time meteorological risk assessment method for low-altitude flight safety, relates to the technical field of flight safety and meteorological information processing, and comprises the following steps: a solar radiation sensitive factor prediction model is constructed, a solar vector is inversed according to latitude and longitude information and attitude parameters acquired by a vehicle in real time, and a radiation risk mapping field covering a flight track area is generated based on the solar vector. The application constructs a radiation risk mapping based on solar vector inversion, realizes forward identification of a thermal disturbance area, and combines a thermal disturbance potential index to realize imaging window scheduling and flight attitude cooperative control. Through multi-channel imaging control driven by a signal-to-noise ratio, frequency hopping sampling and Hadamard exposure modulation, solar radiation interference is effectively inhibited, and image quality is improved. Imaging residual feedback is fed back to the model to realize closed-loop updating, and the safety and sensing reliability of low-altitude flight are comprehensively enhanced.

Description

Technical Field

[0001] This invention relates to the field of flight safety and meteorological information processing technology, specifically to a real-time meteorological risk assessment method for low-altitude flight safety. Background Technology

[0002] "Real-time meteorological risk assessment for low-altitude flight safety" refers to the process of real-time assessment and risk classification of meteorological factors that may affect flight safety during the operation of low-altitude aircraft (such as drones, helicopters, and general aviation aircraft), taking into account the characteristics of low flight altitude and complex and changeable environment. This is achieved by utilizing real-time collected meteorological data (including wind speed, wind direction, temperature, humidity, precipitation, visibility, air pressure, turbulence, etc.) and employing dynamic analysis and risk modeling methods. This allows for the rapid identification of potential hazards (such as low-altitude wind shear, severe convection, localized heavy rain, and dense fog), and the assessment results are fed back to the flight control or monitoring system to help pilots or flight control systems make decisions to avoid risks, make adjustments, or delay operations. Its core significance lies in improving the safety and reliability of low-altitude flight under complex weather conditions through real-time, dynamic quantification and early warning of meteorological risks.

[0003] The existing technology has the following shortcomings:

[0004] In existing technologies, when aircraft perform low-altitude flight missions in high-altitude regions or tropical areas during midday, they are often affected by direct sunlight at high angles. Due to the intense solar radiation, the radiation temperature in local areas of the upper atmosphere can rise sharply in a short period of time, forming a thermal disturbance layer with significant spatial inhomogeneity. In this dynamically changing scenario, the far-infrared monitoring equipment or thermal imaging devices relied upon by the aircraft are easily interfered with by direct sunlight, resulting in a severe attenuation of their signal acquisition capabilities and even the emergence of monitoring blind spots. Existing technologies often rely solely on the anti-interference capabilities of the equipment itself, lacking targeted data correction and dynamic filtering mechanisms, making it difficult to maintain continuous perception of airflow structure under interference conditions. When the aircraft passes through this area, the system will be unable to accurately identify and warn of the updrafts caused by the thermal disturbance layer, which can easily cause the aircraft to lose attitude due to unexpected lift risks, thereby affecting the safety of low-altitude flight.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a real-time meteorological risk assessment method for low-altitude flight safety, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a real-time meteorological risk assessment method for low-altitude flight safety, comprising the following steps:

[0008] A solar radiation sensitivity factor prediction model is constructed. The solar vector is inverted based on the latitude and longitude information and attitude parameters acquired by the spacecraft in real time, and a radiation risk mapping field covering the flight track area is generated based on the solar vector.

[0009] Based on the radiation risk mapping field, a panoramic field-of-view scanning mechanism is initiated to perform high-frequency, multi-angle forward detection, analyze the spatial distribution characteristics of solar radiation in the area in front of the spacecraft, and extract rising thermal disturbance potential indicators for thermal disturbance identification.

[0010] Based on the rising thermal disturbance potential index, the imaging window pre-scheduling operation is performed to adjust the attitude parameters and optical axis orientation parameters of the sensor stabilization platform so that the imaging area avoids areas with strong radiation interference.

[0011] After the imaging window is pre-scheduled, the flight attitude fine-tuning control is executed to adjust the pitch and yaw angles of the aircraft in real time so that the flight path direction is consistent with the imaging optical axis direction.

[0012] Based on image data under the coordinated stability of flight attitude, a real-time imaging quality evaluation matrix is ​​constructed. The signal residuals and image temperature contrast gradient information of multi-channel observation data are fused to generate a signal-to-noise ratio control instruction set for image enhancement.

[0013] Based on the signal-to-noise ratio control instruction set, time-division multiplexing imaging control of the spectral and polarization channels is executed. Multi-band frequency hopping sampling is adopted and the polarization angle is rotated synchronously to drive the micro shutter array to perform exposure modulation according to the Hadamard sequence. The exposure duty cycle is dynamically adjusted according to the solar vector to suppress solar radiation interference during the imaging process. The image residual data is fed back to the solar radiation sensitivity factor prediction model to achieve closed-loop update.

[0014] Preferably, the steps for generating the radiation risk mapping field covering the flight track area are as follows:

[0015] The system acquires the real-time latitude and longitude, flight altitude, pitch angle, yaw angle, and roll angle of the aircraft, marks them with a unified timestamp, and calculates the local time.

[0016] Based on the real-time latitude and longitude of the aircraft, its flight altitude, and local time, the solar vector is calculated and converted from the geographic coordinate system to the direction vector in the aircraft's body coordinate system.

[0017] Based on the structural parameters of the aircraft and the direction of the solar vector, the solar radiation incident angle and radiation intensity response values ​​of the relevant surfaces of the infrared sensor are calculated, and a solar radiation sensitivity factor matrix is ​​constructed in combination with material properties;

[0018] Based on the solar radiation sensitivity factor matrix, a three-dimensional grid region is constructed in front of the spacecraft in the current velocity direction. Radiation interference intensity is predicted and rising thermal disturbance potential is identified for each spatial unit, forming a radiation risk mapping field covering the flight track.

[0019] Preferably, the steps for extracting the rising thermal perturbation potential index are as follows:

[0020] Based on the solar vector direction, the aircraft's heading angle, pitch angle, and roll angle, the radiation interference risk area in the cone-shaped three-dimensional space in front of the aircraft is determined, and this space is divided into scanning units composed of multiple directional vectors.

[0021] The optical sensing device is controlled to scan each directional vector sequentially at a high frequency, and the infrared irradiance, infrared brightness temperature, thermal noise and image sharpness parameters in each direction are collected. The spatial temperature gradient and the rate of change of radiation intensity are also calculated.

[0022] A radiation transition surface is constructed based on the radiation intensity gradient, and regions on the transition surface where both the temperature change rate and the vertical gradient exceed a set threshold are identified and extracted as vertical surge regions of thermal perturbation.

[0023] Calculate the probability of each thermal disturbance vertical surge region intersecting with the aircraft imaging path in the next 3 seconds, and output the optical axis adjustment angle and attitude adjustment suggested parameters.

[0024] Preferably, the imaging window pre-scheduling operation steps are as follows:

[0025] Based on the spatial coordinates, radiation intensity, thermal gradient direction vector, regional influence radius and predicted duration of disturbance contained in the rising thermal disturbance potential index, a forward projection field of view is constructed and a list of shielded areas is generated.

[0026] From the remaining directional field of view outside the shielded area, the imaging retention, safety margin, interference change trend and adjustment energy consumption of each directional point are evaluated at preset elevation and azimuth intervals. The directional point with the highest comprehensive score is selected as the target direction of the imaging window.

[0027] The three-axis angle difference between the current direction of the sensor optical axis and the target direction is calculated and converted into pitch angle, yaw angle and roll angle adjustment amounts. The control parameters are then output to the platform to drive the angle offset control program and complete the imaging window pre-scheduling operation.

[0028] Preferably, after the imaging window is pre-scheduled, the flight attitude fine-tuning control steps are as follows:

[0029] Based on the optical axis direction after the imaging window is pre-scheduled, the optical axis direction is mapped from the aircraft body coordinate system to the geographic coordinate system through the attitude transformation matrix to obtain the spatial expression of the current imaging direction.

[0030] Obtain the current flight path direction vector of the aircraft, calculate the angle difference between the imaging direction and the flight path direction in the vertical and horizontal planes, and generate pitch angle deviation and yaw angle deviation;

[0031] By considering the constraints of the aircraft structure, the control rate boundary, the influence of atmospheric density, and the requirements for interference avoidance, a nonlinear attitude optimization function is constructed to output the optimal attitude target angle.

[0032] The difference between the target attitude angle and the current attitude is converted into control surface deflection, and the elevator and rudder are controlled to smoothly adjust the aircraft's pitch and yaw angles.

[0033] After attitude adjustment is completed, the attitude response results are continuously sampled by the inertial measurement unit, the attitude error is calculated, and proportional-integral-derivative feedback control is performed to ensure that the imaging optical axis direction is consistent with the flight trajectory direction.

[0034] Preferably, the signal-to-noise ratio control instruction set generation steps are as follows:

[0035] Image data is acquired using a multi-channel optical observation device, and each frame of the image is compared with the previous few frames to calculate the signal residual of each pixel, generate a residual matrix, and calculate the mean and variance of the residuals.

[0036] Temperature contrast gradient calculation is performed on each channel image, and the residual matrix and gradient matrix are combined for joint analysis to calculate the local signal-to-noise ratio of each pixel and generate a real-time imaging quality evaluation matrix.

[0037] Based on the evaluation matrix, the channel with the lowest signal-to-noise ratio is identified and a set of signal-to-noise ratio adjustment instructions is generated, including instructions for exposure time adjustment, frame rate modification, image enhancement, channel switching, and image filter weight adjustment, which are used to optimize imaging quality and feed back to the imaging control path.

[0038] Preferably, according to the signal-to-noise ratio modulation instruction set, the time-division multiplexing imaging control steps for the spectral and polarization channels are executed as follows:

[0039] Based on the real-time imaging quality evaluation matrix, the spectral channel and polarization angle with the optimal signal-to-noise ratio are selected, a complete imaging control cycle is set, and the cycle is divided into multiple sub-time periods, with the spectral channel and polarization angle specified for each sub-time period.

[0040] A frequency-hopping sampling mechanism is implemented to dynamically adjust the sampling order based on the spatial angle between the solar vector and the current flight attitude, so as to avoid collecting high-risk channel data when there is strong radiation interference.

[0041] The polarization angle rotation device is controlled to adjust the polarization angle synchronously, and different polarization angles are switched in each sampling cycle to ensure full coverage of imaging requirements for different spectra and polarization states;

[0042] The micro-shutter array is driven to perform exposure modulation according to the Hadamard sequence, and the exposure duty cycle is dynamically adjusted according to the angle between the solar vector and the sensor optical axis, and the exposure time is adjusted in real time to reduce the interference of solar radiation on the image.

[0043] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0044] This invention achieves proactive identification of potential thermal disturbance regions through solar vector inversion and radiation risk mapping; it guides imaging window scheduling and flight attitude fine-tuning using rising thermal disturbance potential indices, achieving precise coordination between imaging field of view and flight attitude; it generates multi-channel imaging control commands by combining signal-to-noise ratio evaluation, dynamically executes frequency-hopping sampling of multispectral and polarization angles, and effectively suppresses image degradation under strong radiation interference by combining micro-shutter array and Hadamard sequence exposure modulation techniques. Finally, the imaging residual information is fed back to the prediction model for adaptive updates, constructing a complete closed-loop mechanism from risk prediction and avoidance control to perception enhancement. This solves the problems of invisible thermal disturbances, severe image distortion, and system response lag in existing technologies, significantly improving the safety, perception stability, and mission reliability of low-altitude vehicles under high radiation disturbance conditions. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0046] Figure 1 This is a flowchart of the real-time meteorological risk assessment method for low-altitude flight safety according to the present invention. Detailed Implementation

[0047] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0048] This invention provides, for example Figure 1 The real-time meteorological risk assessment method for low-altitude flight safety shown includes the following steps:

[0049] A solar radiation sensitivity factor prediction model is constructed. The solar vector is inverted based on the latitude and longitude information and attitude parameters acquired by the spacecraft in real time, and a radiation risk mapping field covering the flight track area is generated based on the solar vector.

[0050] To ensure low-altitude flight safety under strong solar radiation interference, a prediction method based on solar radiation sensitivity factors is proposed. This method combines the aircraft's dynamic state information to construct a radiation risk mapping field covering the flight path in real time, providing a priori basis for subsequent perception control and image scheduling strategies. This method is characterized by strong operability, clear data sources, closed-loop computational path, and precise risk identification. The specific implementation is as follows:

[0051] Real-time flight status data is acquired through a multi-source navigation system installed on the aircraft. Latitude and longitude are collected using high-precision global satellite navigation signals, flight altitude is calculated jointly by a barometric altimeter and GNSS altitude, and attitude parameters, including pitch, yaw, and roll angles, are output by the inertial measurement unit via gyroscopes and accelerometers. The above data is synchronously timestamped and, based on Coordinated Universal Time (UTC) and the aircraft's current geographic time zone, precise local time is obtained.

[0052] Based on the spacecraft's latitude, longitude, altitude, and current time, the solar ephemeris model recommended by the International Astronomical Union is used to calculate the sun's position relative to Earth at the current time. Furthermore, a spatial unit vector is constructed by combining the spacecraft's ground projection point with the sun's direction, yielding a solar vector pointing from the spacecraft's center to the sun's center. This solar vector is first expressed in a geographic coordinate system, then converted into a direction vector in the spacecraft's body coordinate system using an attitude calculation matrix (such as a direction cosine matrix or quaternions). The angles relative to the spacecraft's front, top, and sides are then determined, providing a basic input for subsequent analysis of light-sensitive direction interference.

[0053] After obtaining the direction of the solar vector in the spacecraft's coordinate system, the projection relationship of solar radiation on the external structural surfaces of the spacecraft and the optical axes of the sensors is further analyzed. Based on the structural design drawings of the spacecraft, key surface areas exposed to the sensors on each surface are extracted, including radomes, sensor windows, thruster shells, and support frames. The incident angle of the solar vector is calculated on each surface unit. Combining the reflectivity, transmittance, and absorptivity of each material in the infrared band (e.g., 3–5 μm and 8–12 μm), the energy conversion relationship of solar radiation on each surface is estimated.

[0054] Further, a set of five sensitive factor indicators was constructed: the first is solar irradiance intensity, in watts per square meter; the second is the cosine of the angle between the solar vector and the sensor optical axis, which measures the risk of direct interference; the third is the temperature gradient after heat propagates on the surface of the aircraft, which is calculated by combining heat capacity and thermal conductivity to determine the average surface heating rate; the fourth is the overlapping area of ​​the field of view of reflected light entering the sensor, in square millimeters; and the fifth is the assessment of multi-view shading effect, which calculates the proportion of possible local shading area based on the direction of the solar vector and the structural layout of the aircraft.

[0055] The five factors mentioned above are dynamically calculated based on the current attitude of the aircraft to form a sensitive factor matrix, which is updated once per second to ensure that the trend of solar radiation interference can be reflected in real time during flight, which is different from the low-precision method of using a fixed angle estimation model in the existing technology.

[0056] By combining the aircraft's current orientation with the desired flight path direction, a forward spatial scanning region is defined, forming a three-dimensional spatial region that unfolds in a fan shape with the aircraft body as the vertex. A typical setting is a range of 150 to 500 meters in front of the aircraft's current velocity direction, with a vertical range of ±30 degrees and a horizontal range of ±45 degrees. This fan-shaped region is gridded with a 5-meter spatial resolution, and each grid point is considered a spatial unit to be evaluated.

[0057] For each spatial unit, the angle between its center point's position relative to the sun and the sun's vector direction is calculated at several future time intervals (e.g., 1 second, 3 seconds, 5 seconds). Combined with the previously constructed sensitivity factor response model, the intensity of radiation interference that this point may receive is inverted, and the risk of perception distortion to the spacecraft's infrared observations is further calculated. The assessment results for each spatial unit are categorized by risk level (low, medium, relatively high, high, extremely high) to form a spatial distribution map.

[0058] To improve the timeliness of risk assessment results, this step implements a dual-time-window fusion mechanism. On the one hand, it records the spatial interference intensity at the current time point, and on the other hand, it extrapolates the position of each cell on the flight path in the next 3 seconds based on the aircraft's current speed, dynamically projecting it onto the direction of the solar vector to form a time-sensitive interference prediction model in the forward-looking direction, further enhancing the forward-looking identification capability of future interference situations.

[0059] The generated radiation risk map field is projected onto the effective field of view of the aircraft's optical imaging sensor. The analysis focuses on identifying high-risk areas that might enter the imaging optical axis path under the current attitude and anticipated future attitude change trends. For unit points marked as high-risk and extremely high-risk in the map field, their position relative to the aircraft body, angle with the sensor's optical axis, interference direction distribution range, and interference energy level are extracted.

[0060] These characteristic values ​​are combined as indicators of high interference risk, and combined with the response rate and control boundaries of the aircraft's attitude control system (e.g., pitch control range ±20 degrees, yaw control rate ≤30 degrees / second) to determine whether the aircraft has the ability to complete attitude avoidance before the high-risk area appears. Suggested attitude correction angles are output for avoidable areas, and unavoidable areas are marked as points with strong interference potential.

[0061] Further analysis of the spatial continuity of temperature gradient changes and optical interference boundaries using sensitive factors identifies regions with drastically changing edges and labels them as high-potential regions for rising thermal disturbances. These regions exhibit rapid temperature jumps and a high probability of turbulence. By marking their locations and eigenvalues, the results are used to guide subsequent parameter adjustments in imaging window pre-scheduling, flight attitude fine-tuning strategies, and image signal-to-noise ratio control steps.

[0062] Based on the radiation risk mapping field, a panoramic field-of-view scanning mechanism is initiated to perform high-frequency, multi-angle forward detection, analyze the spatial distribution characteristics of solar radiation in the area in front of the spacecraft, and extract rising thermal disturbance potential indicators for thermal disturbance identification.

[0063] Building upon the existing three-dimensional radiation risk mapping field covering the flight path, this paper proposes a method for detecting spatial radiation structures and extracting thermal disturbance potential indicators based on full-field-of-view scanning to enhance the forward spatial risk perception capability of aircraft under conditions of high-intensity solar radiation interference. This method dynamically analyzes the radiation distribution in the airspace ahead of the aircraft by controlling the sensor's scanning attitude, detection frequency, and data processing flow, thereby obtaining forward thermal disturbance risk perception results with low time lag, high spatial resolution, and high recognition efficiency.

[0064] Based on the acquired solar vector data (i.e., the direction vector of the sun relative to the spacecraft's coordinate system), and the spacecraft's current heading, pitch, and roll angles, areas where solar radiation may interfere with the forward field of view are identified. Using the spacecraft's forward inertial direction as the axis, a conical three-dimensional scanning area is defined, with a horizontal coverage angle of 120 degrees (60 degrees to the left and right) and a vertical coverage angle of 90 degrees (45 degrees above and 45 degrees below). The maximum scanning depth is set to 500 meters based on the spacecraft's speed and is automatically adjusted with the flight speed to meet the spatial distribution requirements within the forward prediction time window (e.g., 5 seconds).

[0065] The three-dimensional conical space is divided into multiple scanning units with fixed pose directions. Each scanning unit is defined as a specific direction vector, with an angular interval of 5 degrees horizontally and 5 degrees vertically, thus forming a spatial scanning point array with uniform density. Commands for all scanning directions are loaded into the attitude control device, driving the optical sensing equipment to point in each direction at a scanning frequency of 20 Hz (completing one full field-of-view scan per second) and complete information sampling. This step completely abandons the "fixed-angle observation" method of traditional thermal imaging equipment. Through real-time scanning and continuous updating of a large field of view, it achieves the active acquisition capability of full-space information and constructs a perception framework coupled in real-time with the flight mission.

[0066] After the field of view is constructed, infrared radiation is sampled sequentially in each scanning direction. The infrared sampling device includes a photosensitive sensor, a digital signal conversion circuit, and an environmental calibration unit, with the operating wavelength covering two atmospheric window ranges: 3–5 micrometers and 8–12 micrometers. In each direction, irradiance, infrared brightness temperature, grayscale image intensity, thermal noise figure, and image sharpness distribution are continuously acquired within 1 millisecond. Simultaneously, samples are taken from the adjacent 5-degree regions above and below that direction to obtain the temperature gradient changes in that spatial direction.

[0067] Subsequently, the acquired data is filtered in the time dimension to remove short-period disturbances caused by high-frequency attitude disturbances of the aircraft. Simultaneously, in the spatial dimension, the finite difference method is used to calculate the radiation intensity gradient along adjacent sampling directions to identify local non-uniform radiation distribution zones. These strong gradient regions represent the trend lines of rapid spatial transfer of thermal disturbances, and their distribution can be reconstructed in a three-dimensional coordinate system as a "radiative transition surface," which has an irregular curved structure and closely matches the boundary of the actual radiation disturbance. Compared with traditional thermal imaging methods that rely on "temperature difference blocks" or "image bright spots," this method possesses higher structural integrity and spatial coherence, enabling more accurate prediction of the spatial location of disturbance sources.

[0068] In the radiation spatial reconstruction map obtained in the second step, based on the distribution characteristics of the radiation gradient field, regions with a temperature change rate exceeding 10 degrees Celsius per second and a spatial thermal energy density gradient greater than 100 watts per cubic meter are selected as the first layer of screening criteria to identify unstable regions with highly concentrated thermal energy that may trigger updrafts. Further calculation of the vertical radiation directionality of these regions is then performed. If the vertical gradient of a region is greater than the horizontal gradient by more than 20%, it indicates that the hotspot region has an upward trend and is defined as a "vertical surge zone of thermal disturbance".

[0069] A "thermal disturbance potential index set" is established for all vertical surge areas, including the following five explicit quantitative parameters: (1) radiation variation amplitude, in watts per square meter; (2) local spatial thermal gradient distribution, in watts per cubic meter; (3) radius of the disturbance area, in meters; (4) main propagation direction vector of thermal disturbance, expressed in body coordinate system; (5) probability of intersection between the disturbance area and the flight path in the next 3 seconds, calculating whether the hot spot will enter the forward-looking window area of ​​the aircraft. Once the above indicators are formed, they are stored as structured data and used as active avoidance decision parameters in flight path perception. Compared with the existing single-dimensional judgment method that relies on image mutation or fluctuation amplitude, this method has four-dimensional judgment criteria: structure, directionality, continuity, and predictability, which improves the reliability of risk judgment.

[0070] Multiple potential thermal disturbance regions are extracted and mapped onto the aircraft's current navigation coordinate system. It is then predicted whether these regions will coincide with the sensor's optical axis path within several future time windows (e.g., 1 second, 3 seconds, 5 seconds). A vector projection method is used to determine whether the main propagation direction of the thermal disturbance is orthogonal to or opposite to the aircraft's imaging direction. If the angle is within 15 degrees, it is considered to be at high risk of interference conflict.

[0071] Based on this result, two key feedback parameters are generated: (1) a suggested optical axis adjustment angle, quantified as the specific angle to be offset up, down, left, and right in degrees, used to adjust the direction of the sensor imaging window to avoid entering the high-risk area of ​​thermal disturbance; (2) a suggested attitude adjustment command, quantified as pitch and yaw angle correction values, combined with flight control constraints and control response speed, as a suggested scheme for fine-tuning the flight path. The above parameters are transmitted to the control command execution link, and finally the prior information input for imaging window pre-scheduling and attitude synchronization adjustment is completed.

[0072] Based on the rising thermal disturbance potential index, the imaging window pre-scheduling operation is performed to adjust the attitude parameters and optical axis orientation parameters of the sensor stabilization platform so that the imaging area avoids areas with strong radiation interference.

[0073] To ensure the quality of data acquired by the spacecraft during imaging missions under conditions of strong solar radiation interference, and to avoid problems such as saturation, polarization disturbances, temperature distortion, or loss of target features in the images, real-time pre-scheduling of the imaging window is necessary after extracting the thermal disturbance potential index. Pre-scheduling refers to adjusting the attitude parameters of the sensor stabilization platform and the optical axis direction of the sensor in advance based on the predicted location and intensity of thermal disturbances. This ensures that the field of view about to enter the imaging window avoids high-interference areas, thereby maintaining the stability and continuity of the imaging.

[0074] This pre-scheduling method aims to achieve high timeliness, high spatial resolution, and high-precision steering control, and is accomplished through the following steps.

[0075] The thermal disturbance potential index is generated from the preceding steps and includes five parameters: first, the spatial coordinates of the disturbance area, represented in the aircraft's coordinate system, including lateral, longitudinal, and altitude difference values; second, the radiation intensity value, in watts per square meter, used to reflect the degree of disturbance; third, the thermal gradient direction vector, a unit vector describing the main propagation direction of radiated energy in the disturbance area; fourth, the radius of influence of the area, in meters, calculated based on the radiation intensity and spatial diffusion rate; and fifth, the predicted duration of the disturbance, in seconds, used to determine whether the disturbance source is a transient or continuous disturbance.

[0076] Based on the five parameters mentioned above, a forward-projected field of view is constructed using the aircraft's current heading vector. Within this space, the angle between the spatial coordinates of each potential disturbance source and the aircraft's forward trajectory is used to determine whether the thermal disturbance will enter the sensor imaging window within a set time window (e.g., 3 seconds). The sensor imaging window has a forward-facing cone shape, with a typical horizontal viewing angle of ±25 degrees, a vertical viewing angle of ±20 degrees, and a scanning depth range of 200 to 500 meters.

[0077] All interference areas meeting the criteria of "entering the field of view + interference level of high or extremely high" are included in the interference screening model. A shielded area index list is constructed using solid angle units. This list records the direction of each interference area in the body coordinate system (labeled with elevation and azimuth angles), estimated arrival time, estimated interference intensity level, and suggested avoidance angle and direction range (e.g., elevation angle shifted downwards by 5 to 15 degrees, azimuth angle shifted to the left by 10 to 20 degrees). This index is sorted chronologically and refreshed every 1 second, ensuring real-time scheduling response. Unlike existing technologies that use temperature difference images to filter out shielded areas, this method has three-dimensional spatial directionality and temporal prediction capabilities, enabling effective advance imaging adjustments.

[0078] When an aircraft performs an observation mission, its target area has a clear directionality. To ensure that the imaging mission is not affected by interference while maintaining target visibility, it is necessary to select the optimal imaging window direction in the non-interference area and generate stable platform attitude control variables for it.

[0079] The specific operation is as follows: First, in the directional space excluding all areas marked as shielded, a remaining usable field of view is constructed at 1-degree elevation and 1-degree azimuth intervals. For each directional point, the following four evaluations are performed: 1) the angle between the directional point and the imaging target direction, used to determine the field-of-view target retention; 2) the distance between the directional point and the boundary of the nearest interference area, used to assess the safety margin; 3) the trend of the predicted interference value in the directional point over the next 3 seconds, used to determine the sustained safety; and 4) the total rotation angle required to adjust from the current sensor direction to the target direction, used to assess the adjustment energy consumption and speed.

[0080] The evaluation scores of all directional points are normalized to between 0 and 1. After weighted calculation, the directional point with the highest score is selected as the target for adjustment of the imaging window in this round. The elevation and azimuth angles of this directional point are the target directions for adjusting the sensor's optical axis.

[0081] Subsequently, the required angular offset from the current attitude of the stabilizing platform to the target attitude is calculated. Starting from the current sensor optical axis direction, the three-axis differences between the target direction and the current direction are calculated, representing the pitch, yaw, and roll angle adjustments (in degrees). For example: if the current pitch angle is 10 degrees and the target direction is 5 degrees, the pitch angle is adjusted down by 5 degrees; if the current azimuth angle is +20 degrees and the target direction is +5 degrees, the yaw angle is adjusted left by 15 degrees; if the stabilizing platform needs to compensate for sensor tilt in the roll direction, this is converted into the required roll angle. All these adjustments are limited to the platform's allowable load range (e.g., maximum pitch angle ±30 degrees, maximum adjustment rate 2 degrees per second) and output to the execution unit in the form of an angular velocity control curve.

[0082] This optimized path planning not only ensures that the imaging direction avoids high-risk areas of thermal disturbance, but also takes into account the continuity of the flight target and the platform's responsiveness. Compared with fixed field-of-view adjustment mechanisms or unidirectional avoidance strategies, it has more dynamic, intelligent and multi-dimensional balanced characteristics.

[0083] The calculated three-axis angle adjustment is used as the control input parameter and sent in real time to the actuators on the aircraft via the data link, driving the stabilization platform to initiate the angle offset control program. According to the attitude adjustment command, the platform controls the pitch axis, yaw axis and roll correction mechanism respectively, so that the sensor slowly turns to the target direction at a set angular velocity. The whole process maintains a smooth curve adjustment without oscillation or jump, so as to avoid image blurring or inertial shock caused by rapid rotation.

[0084] After the sensor completes angle adjustment, a high-speed image sampling is immediately performed. Three methods—multi-channel infrared imaging, thermal layer brightness analysis, and target sharpness assessment—are used to quickly determine the interference intensity under the new imaging direction. If the new image has no strong radiation saturation areas, the thermal gradient change is within 3 degrees Celsius, and the target edge sharpness exceeds 90% of the reference threshold, the imaging window pre-scheduling is confirmed to be successful. Otherwise, the current imaging direction information is transmitted back to reassess whether the interference area has moved or been misjudged, and the direction optimization step is restarted to begin the next round of fine-tuning.

[0085] Furthermore, to avoid unstable imaging caused by the imaging window constantly switching during continuous flight, this implementation introduces a direction stabilization mechanism. When the interference area ahead does not change significantly, the system will lock the current imaging window direction for more than 3 seconds, prohibiting the platform from freely deflecting, thus improving imaging continuity. In areas where interference sources frequently change, a stable direction is prioritized, i.e., the direction with no significant disturbance trend in the future prediction time window is selected as the candidate imaging window.

[0086] This closed-loop execution and verification method enables the imaging window pre-scheduling to form a complete link from "prediction - command - adjustment - verification", ensuring that all operations are based on dynamic responses to actual flight conditions. This eliminates the incompatibility and lag problems caused by the traditional static imaging window arrangement method, laying a stable and reliable foundation for subsequent attitude coordination and imaging signal-to-noise ratio optimization.

[0087] After the imaging window is pre-scheduled, the flight attitude fine-tuning control is executed to adjust the pitch and yaw angles of the aircraft in real time so that the flight path direction is consistent with the imaging optical axis direction.

[0088] After completing the pre-scheduling of the imaging window, the sensor's optical axis has been adjusted to a new observation direction that avoids areas of strong radiation interference. To ensure that the imaging direction is consistent with the aircraft's trajectory, thereby improving target tracking stability and image continuity, the aircraft's pitch and yaw angles must be adjusted in real time to align the aircraft's forward motion trajectory as closely as possible with the imaging optical axis. This step achieves high-precision coordination between the imaging optical axis and the flight trajectory by implementing fine-tuning control of the flight attitude step by step.

[0089] After the aircraft completes the imaging window adjustment, the direction value of the current optical axis in the aircraft's body coordinate system needs to be obtained from the angle sensor of the stabilized platform. This direction is usually represented as a unit vector, containing three components: the x-axis represents the forward direction, the y-axis represents the lateral direction, and the z-axis represents the vertical direction. The optical axis direction vector in this body coordinate system is mapped to the geographic coordinate system through an attitude transformation matrix, thereby obtaining the true direction expression of the imaging direction in three-dimensional space.

[0090] Simultaneously, the current heading velocity vector of the aircraft is read from the flight control unit and decomposed into a unit vector form as the aircraft's trajectory direction vector. This trajectory direction should comprehensively consider the current flight speed, heading angle, wind speed vector correction, and sideslip angle correction results to ensure that the direction accurately reflects the aircraft's propulsion direction.

[0091] Subsequently, the angle between the imaging optical axis and the flight path direction is calculated using vector dot product, and the pitch and yaw angle deviations are calculated in both the vertical and horizontal projection planes. The goal of this step is to determine the attitude angles that the aircraft needs to adjust to achieve directional alignment.

[0092] The calculated pitch and yaw deviations are used as input parameters, and the target values ​​for attitude adjustment are constructed by combining them with the adjustable attitude angle range of the aircraft. Assuming the current pitch angle is 10 degrees and the yaw angle is 15 degrees, and according to the optical axis alignment requirements, it needs to be adjusted to 7 degrees pitch and 10 degrees yaw, then the desired attitude adjustment is to adjust the pitch down by 3 degrees and the yaw left by 5 degrees.

[0093] To ensure the feasibility and safety of this adjustment, several constraints need to be introduced to establish the optimization function: Constraint 1 is the maximum and minimum pitch angles allowed by the current structure of the aircraft (e.g., +25 degrees to -10 degrees); Constraint 2 is the maximum allowable yaw rate (e.g., not exceeding 25 degrees per second); Constraint 3 is the influence factor of the current flight altitude and atmospheric density on the attitude control sensitivity; Constraint 4 is whether the adjustment will cause the aircraft to re-enter the high-radiative thermal disturbance region; Constraint 5 is the compatibility between the adjustment time and the observation mission time window.

[0094] By incorporating the aforementioned constraints into the nonlinear multi-objective optimization function, the adjustment ranges of pitch and yaw angles are restricted and filtered to output the optimal attitude target angle that meets the requirements of flight safety and perception.

[0095] The difference between the target attitude angle value and the current attitude value is calculated to obtain the actual attitude adjustment amount that needs to be performed. Taking a fixed-wing aircraft as an example, the pitch angle is usually adjusted by the elevator, and the yaw angle is adjusted by the rudder. According to the aerodynamic response model of the aircraft, if the current pitch angle is 10 degrees and needs to be adjusted to 7 degrees, then the elevator needs to be deflected down by a certain number of degrees. For example, if a unit rudder surface deflection of 1 degree can cause a pitch change of 0.5 degrees, then a pitch reduction of 3 degrees corresponds to a elevator deflection of 6 degrees.

[0096] The same analysis is performed on the yaw angle. If the yaw difference is 5 degrees, the rudder needs to be deflected by a corresponding amount. For example, if a 1-degree deflection of the rudder results in a 0.25-degree change in the yaw angle, then a 20-degree deflection is required. These values ​​are then input into the flight control execution path in a control command format to drive the aircraft to adjust its attitude.

[0097] During this process, an upper limit for angular velocity should be set (e.g., no more than 5 degrees per second), and a smooth curve should be used for control input interpolation to avoid instantaneous changes and prevent overshoot or chattering. At the same time, aerodynamic hysteresis should be considered to pre-compensate for control response delay, so that the aircraft attitude can respond promptly after the command input and approach the target attitude value.

[0098] During attitude adjustment, the aircraft's real-time attitude angles are sampled at a high frequency (e.g., updated 100 times per second). The current pitch and yaw angles are obtained through the inertial measurement unit and attitude calculation algorithm, and compared with the desired attitude values ​​to calculate the attitude error. If the error is within a set threshold (e.g., less than 1 degree), the adjustment is considered successful; otherwise, it enters the secondary adjustment stage.

[0099] To avoid attitude adjustment failure due to external disturbances (such as wind field changes or gyroscope drift), a proportional-integral-differential algorithm is used to perform secondary correction on the control input and gradually reduce the deviation until the attitude angle enters the stable range. At this point, the angle between the imaging optical axis and the flight trajectory direction is calculated again to verify whether the orientation is consistent after attitude adjustment.

[0100] If there is still a significant angle in the direction, but the attitude cannot be adjusted further (such as reaching the structural limits or safety boundaries of the aircraft), this information is fed back to the imaging window scheduling logic to recalculate the acceptable direction range of the optical axis. Fine-tuning of the optical axis is used instead of flight attitude adjustment to ensure that the overall observation performance is maintained at a high level of stability.

[0101] After completing the aforementioned attitude fine-tuning, the current aircraft attitude, track direction, and imaging optical axis direction are used as the new flight mission reference benchmarks and recorded as the attitude-track-imaging coupled state. In this state, the aircraft's motion direction is basically consistent with the observation direction, ensuring subsequent imaging continuity and image clarity.

[0102] Simultaneously, a dynamic stability maintenance mechanism is set under this baseline state. Without new thermal disturbances or sudden changes in the mission target direction, the system prohibits unnecessary attitude changes and optical axis disturbances, allowing the aircraft to enter a stable cruise imaging phase. If the duration of the stable attitude state exceeds a preset threshold (e.g., 5 seconds), this stable state is used as the initial input for the next stage of trajectory calculation and disturbance prediction, achieving closed-loop linkage between perception, control, and imaging.

[0103] By employing the above methods, the aircraft can maintain a high degree of consistency between the sensor imaging direction and the flight propulsion direction while avoiding interference areas, thereby achieving highly stable image acquisition and minimizing image blurring, jitter, or target offset caused by attitude deviation. This effectively distinguishes it from existing control methods that decouple flight and imaging, have lag in attitude response, and frequently deviate from the imaging path.

[0104] Based on image data under the coordinated stability of flight attitude, a real-time imaging quality evaluation matrix is ​​constructed. The signal residuals and image temperature contrast gradient information of multi-channel observation data are fused to generate a signal-to-noise ratio control instruction set for image enhancement.

[0105] After the flight attitude and imaging optical axis are coordinated and finely adjusted, the spacecraft enters a stable observation state. During this stage, the spacecraft's sensors can stably acquire multi-channel image data along the flight path. To ensure the effectiveness and stability of the imaging data under radiation interference, this step constructs a real-time imaging quality evaluation matrix, analyzes the signal residuals and temperature contrast gradients in the image, and generates a quantifiable signal-to-noise ratio (SNR) control instruction set to guide subsequent dynamic adjustment of imaging parameters and execution of image enhancement strategies. The specific implementation method is as follows:

[0106] With the aircraft in a stable attitude, a multi-channel optical observation device mounted on a stable platform continuously acquires images across multiple spectral channels, including mid-infrared images, long-wave thermal radiation images, and visible light high dynamic range images. The image sampling frequency is set to 20 frames per second, with each frame's timestamp accurate to the millisecond level, ensuring complete temporal alignment between different channels. Each image undergoes geometric calibration and pixel registration to achieve spatial alignment, ensuring that any pixel points to the same ground position in different channels.

[0107] The acquired raw image sequence is processed through image alignment and time-domain filtering to extract its residual signal. The residual signal is calculated as follows: using the past 5 consecutive frames as a time window, a weighted average method is used to generate a predicted image for that channel as a background reference image. The current image is subtracted from the background image to obtain the brightness deviation value of each pixel in the current frame, and its absolute value is used as the signal residual value of that pixel.

[0108] The residual values ​​of each pixel in the entire image are used to form a two-dimensional residual matrix. The mean and variance of this matrix are calculated to obtain the overall observation residual signal-to-noise level index for the current image in that channel. This process is repeated to calculate the residual matrix and statistical index for each channel. Finally, a preliminary observation residual array is formed, with the channel as the dimension, time as the horizontal axis, and residual statistical values ​​as the elements.

[0109] This step differs from existing technologies that judge the degree of imaging interference by overall brightness changes. It adopts a time-series modeling + spatial residual statistics approach, which can accurately describe noise behavior at the level of image details and improve the sensitivity of abnormal area identification.

[0110] Based on the residual extraction, temperature contrast gradient calculation is performed on each channel image to measure the image's structural sharpness and edge feature representation capability. Taking the infrared thermal imaging channel image as an example, the gray value of each pixel represents the infrared radiation intensity of its corresponding surface target, and the gray value difference reflects the temperature gradient.

[0111] To address this, an image gradient algorithm is employed to calculate the first-order difference between pixels in both the horizontal and vertical directions. This involves using the grayscale differences between the current pixel and its adjacent pixels (top, bottom, left, and right) to generate the horizontal and vertical gradients for that point. The gradient values ​​from both directions are then combined to form the total gradient intensity for that pixel, expressed in pixels as grayscale differences.

[0112] By statistically analyzing the gradient values ​​of all pixels in the entire image, the following key indicators are extracted: (1) average gradient value, used to evaluate the overall contrast level; (2) maximum gradient value and corresponding position, used to identify possible target edges; (3) standard deviation of gradient value, used to reflect the complexity of image structure. For multi-channel images, this process is repeated to calculate the gradient distribution matrix of each channel and store it as a data structure paired with the residual matrix.

[0113] By jointly analyzing the residual matrix and the gradient matrix, the noise residual intensity and contrast gradient value are compared at the same pixel location to construct a local signal-to-noise feature map. That is, the degree of difference between the local target signal and the background is used as the "signal intensity" and the local noise residual is used as the "interference intensity". The local signal-to-noise ratio is calculated for each pixel.

[0114] This step, by introducing gradient distribution as a measure of image structure signal, achieves a technical transition from "grayscale contrast" to "temperature gradient structure," significantly improving the spatial resolution and realism of image quality assessment and overcoming the limitation of existing methods that neglect the sharpness of target boundaries.

[0115] The overall image residual statistics for each channel are combined with the average gradient intensity value of the corresponding channel. Based on the "signal-to-interference" ratio principle, the average signal-to-noise ratio of the entire image at the current time point is calculated, with the unit being "effective gray level / noise level". The signal-to-noise ratio value of each channel at each time point is filled into a two-dimensional table, with rows representing different time points and columns representing different channels, forming a "real-time imaging quality evaluation matrix".

[0116] Based on this matrix, the channel with the lowest signal-to-noise ratio (SNR) at the current time point and its changing trend are extracted as preliminary early warning signals of image quality degradation. If the SNR of a certain channel continuously decreases by more than 10% over multiple time windows, while its mean residual increases by more than 30%, the channel is determined to be in a high-interference state. The causes of the SNR decrease are then analyzed by combining the current aircraft attitude change parameters with the thermal disturbance prediction region in the radiation risk mapping field.

[0117] After completing the signal-to-noise ratio (SNR) status assessment, a SNR adjustment instruction set is generated, including the following five categories of instructions:

[0118] Exposure time adjustment command: Based on the signal-to-noise ratio change trend, extend or shorten the infrared shutter time in microseconds, with the control range set to ±30% of the current exposure time.

[0119] Frame rate modification instruction: In the case of a low signal-to-noise ratio scene, temporarily increase the image sampling frame rate, for example, from 20 frames per second to 30 frames per second, to increase the amount of redundant information for subsequent fusion enhancement.

[0120] Image enhancement parameter instructions: dynamically adjust the image contrast stretching coefficient and nonlinear mapping function parameters to enhance the image visibility in low-contrast areas. The specific values ​​are obtained by fitting the extreme points of the gradient distribution.

[0121] Channel switching command: When the signal-to-noise ratio of a certain channel is continuously lower than the set threshold (e.g., 3:1), the backup observation channel will be automatically activated as the main imaging channel, such as switching from long-wave infrared to mid-wave infrared.

[0122] Image filter weight adjustment command: Enhance the temporal filter weight for areas with large residual fluctuations to reduce the introduction of high-frequency noise. The specific filter weight is automatically set according to the standard deviation of the residual change.

[0123] This instruction set is immediately submitted to the imaging control path after generation, fused with the current sensor state, and executed. After the next frame of image is acquired, its effect on signal-to-noise ratio control is checked to see if it meets the standard. If it does not meet the standard, parameters are rolled back, strategies are adjusted, or the control process is restarted, forming a closed-loop feedback control logic for quality assessment and control.

[0124] According to the signal-to-noise ratio control instruction set, time-division multiplexing imaging control of spectral and polarization channels is executed. Multi-band frequency hopping sampling method is adopted and polarization angle is rotated synchronously to drive the micro shutter array to perform exposure modulation according to the Hadamard sequence. The exposure duty cycle is dynamically adjusted according to the solar vector to suppress solar radiation interference during the imaging process and feed the image residual data back to the solar radiation sensitivity factor prediction model to achieve closed-loop update.

[0125] After constructing the signal-to-noise ratio evaluation matrix and generating the imaging parameter adjustment instruction set, to ensure high-quality image data acquisition even under strong solar radiation interference, a time-division multiplexing approach for spectral and polarization channels is employed for imaging control. This is combined with a frequency-hopping sampling strategy, dynamic polarization angle rotation control, a micro-shutter array exposure modulation mechanism, and a solar vector-driven exposure duty cycle adjustment method to suppress incident solar radiation interference. Furthermore, the final image residual information is fed back to the prediction model, achieving dynamic closed-loop update control throughout the entire process. The specific steps are as follows:

[0126] Based on the real-time imaging quality assessment matrix established in the previous stage, the preferred sequence of spectral channels and the recommended set of polarization angle parameters with the best current image quality are obtained. It is assumed that under the current radiation risk distribution, the mid-wave infrared channel and the long-wave infrared channel are the two channels with the highest signal-to-noise ratio, and the polarization angles are represented by 0°, 45°, 90°, and 135°.

[0127] Based on this, a complete imaging control cycle (e.g., 2 seconds) is set, and this cycle is divided into multiple fixed sub-segments (e.g., 8 sub-segments, each 250 milliseconds). Each sub-segment is explicitly assigned a spectral channel and a polarization angle. For example, the first segment acquires a mid-wave infrared +0° polarized image, the second segment acquires a mid-wave infrared +45° image, the third segment acquires a long-wave infrared +90° image, and so on, arranged sequentially to ensure that every channel-polarization angle combination is covered, and to control its repetition frequency and sampling interval.

[0128] This setting not only ensures that the imaging signal covers the entire band and all polarization angles, but also provides redundant dimensions for subsequent data enhancement and image reconstruction, enabling higher fault tolerance and anti-interference capabilities in complex scenarios.

[0129] To avoid strong solar radiation interference in specific wavelength bands during specific time periods, this step introduces a frequency-hopping sampling mechanism based on the preset channel sampling order. This mechanism uses the spatial angle parameter between the solar vector and the current flight attitude to weight the risk of interference to each spectral channel. For example, when the angle between the solar vector and the field of view of the mid-wave infrared channel is less than 25 degrees, the system classifies it as a strong interference state, postpones the originally planned mid-wave infrared channel sampling task, and instead performs long-wave infrared or near-infrared channel sampling ahead of schedule.

[0130] The sampling jump priority is dynamically planned using the current imaging quality and signal-to-noise ratio as weight values ​​to ensure that there is no over-reliance on a certain channel within the same period, and that data from that channel is not collected when interference is at its maximum, thereby improving the overall image stability.

[0131] Simultaneously, polarization angle synchronous rotation control is performed with each channel switch. The polarization angle setting range is 0° to 180°, and it rotates in steps of 45°, maintaining each angle until the end of the corresponding sub-period. The polarization control device (such as an adjustable LCD polarizer) adjusts the voltage according to the set angle to achieve rapid and stable switching of the target polarization state.

[0132] By controlling the frequency hopping spectrum and polarization angle together, multi-dimensional image sets can be obtained in different combinations of spectrum and polarization, thus achieving spatial decoupling of the directional interference of solar radiation and improving image recognition and structure reconstruction capabilities.

[0133] Based on simultaneous multi-channel and multi-polarization angle imaging, a micro-shutter array is used for high-frequency exposure control to further suppress strong transient illumination disturbances. This shutter array consists of multiple electronically controlled liquid crystal shutters, with each shutter controlling the exposure state of an independent image pixel area, possessing high temporal accuracy and response speed (response time less than 100 microseconds).

[0134] All shutters are time-synchronized according to a Hadamard sequence matrix. The Hadamard matrix is ​​a square matrix composed of orthogonal ±1 squares, where each row corresponds to the shutter's open / closed state during one exposure cycle. For example, a 16×16 Hadamard matrix is ​​set, and the on / off control corresponding to the matrix rows is executed sequentially in the 16 shutter units. After 16 consecutive executions, 16 image subframes with alternating disturbed and undisturbed areas are acquired.

[0135] These image subframes are decoded through the mathematical inversion process of Hadamard transform, achieving noise reduction and suppression of strong radiation interference signals and structural image restoration. This processing not only completes the active separation of interference signals during the imaging stage, but also reduces the dependence of post-processing on image quality.

[0136] To further enhance the modulation effect, the exposure duty cycle is adjusted in real time based on the angle between the solar vector and the sensor's optical axis. If the current angle between the solar vector direction and the optical axis is less than 20 degrees, indicating a high probability of direct sunlight, the system shortens the exposure time to 30% of the original exposure time and extends the shutter closure state to suppress the instantaneous impact of strong light; if the angle is greater than 60 degrees, the system gradually restores the original exposure duty cycle.

[0137] This method combines directional adjustment and time duty cycle control to achieve an imaging strategy of "weakening the sun's rays and enhancing the sun's shadow," effectively improving the dynamic illumination adaptability under complex spatial attitude changes.

[0138] After the imaging task is completed, residual extraction processing is performed on the image under each channel and polarization angle combination. The specific method is as follows: in each image frame, it is compared with the image under similar posture in history to extract the residual image of light intensity change; then, by statistically analyzing the mean and standard deviation of the residual image, radiation interference response data is formed.

[0139] The above data was uploaded with tags including spectral channel ID, polarization angle, shooting time, flight attitude parameters, solar vector direction, image frame number, and corresponding residual signal intensity value. This data set was then input into the solar radiation sensitivity factor prediction model for weight correction and factor reconstruction.

[0140] If the model predicts a low solar radiation risk value for a certain area, but the actual image residual is higher, the model will increase the weight parameters for that area and update the surface reflectance, local atmospheric transmittance, or solar altitude angle sensitivity factor to make the prediction more accurate in the next imaging. This feedback cycle is controlled to be completed once every 5 seconds to quickly respond to attitude and environmental changes during continuous flight.

[0141] Ultimately, a closed-loop process of "prediction-acquisition-identification-feedback-update" is formed. Unlike traditional imaging systems that operate with fixed parameters, it has the ability to quickly adapt to dynamic environments and a mechanism to improve prediction accuracy. It is a key technology to ensure flight mission safety and stable imaging quality.

[0142] The proposed real-time meteorological risk assessment method for low-altitude flight safety enables dynamic modeling, proactive avoidance, and closed-loop control of thermal disturbances caused by high-angle direct sunlight during flight missions, significantly improving the perception continuity and imaging stability of aircraft under complex lighting conditions. This method achieves forward-looking identification of potential thermal disturbance regions through solar vector inversion and radiation risk mapping; it guides imaging window scheduling and flight attitude fine-tuning using rising thermal disturbance potential indices, achieving precise coordination between imaging field of view and flight attitude; it generates multi-channel imaging control commands based on signal-to-noise ratio evaluation, dynamically executes multispectral and polarization angle frequency-hopping sampling, and effectively suppresses image degradation under strong radiation interference by combining micro-shutter arrays and Hadamard sequence exposure modulation techniques. Finally, the imaging residual information is fed back to the prediction model for adaptive updates, constructing a complete closed-loop mechanism from risk prediction and avoidance control to perception enhancement. This solves the problems of invisible thermal disturbances, severe image distortion, and system response lag in existing technologies, significantly improving the safety, perception stability, and mission reliability of low-altitude aircraft under high radiation disturbance conditions.

[0143] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A real-time meteorological risk assessment method for low-altitude flight safety, characterized in that, Includes the following steps: A solar radiation sensitivity factor prediction model is constructed. The solar vector is inverted based on the latitude and longitude information and attitude parameters acquired by the spacecraft in real time, and a radiation risk mapping field covering the flight track area is generated based on the solar vector. Based on the radiation risk mapping field, a panoramic field-of-view scanning mechanism is initiated to perform high-frequency, multi-angle forward detection, analyze the spatial distribution characteristics of solar radiation in the area in front of the spacecraft, and extract rising thermal disturbance potential indicators for thermal disturbance identification. Based on the rising thermal disturbance potential index, the imaging window pre-scheduling operation is performed to adjust the attitude parameters and optical axis orientation parameters of the sensor stabilization platform so that the imaging area avoids areas with strong radiation interference. After the imaging window is pre-scheduled, the flight attitude fine-tuning control is executed to adjust the pitch and yaw angles of the aircraft in real time so that the flight path direction is consistent with the imaging optical axis direction. Based on image data under the coordinated stability of flight attitude, a real-time imaging quality evaluation matrix is ​​constructed. The signal residuals and image temperature contrast gradient information of multi-channel observation data are fused to generate a signal-to-noise ratio control instruction set for image enhancement. Based on the signal-to-noise ratio control instruction set, time-division multiplexing imaging control of the spectral and polarization channels is executed. Multi-band frequency hopping sampling is adopted and the polarization angle is rotated synchronously to drive the micro shutter array to perform exposure modulation according to the Hadamard sequence. The exposure duty cycle is dynamically adjusted according to the solar vector to suppress solar radiation interference during the imaging process. The image residual data is fed back to the solar radiation sensitivity factor prediction model to achieve closed-loop update.

2. The real-time meteorological risk assessment method for low-altitude flight safety according to claim 1, characterized in that, The steps for generating a radiation risk mapping field covering the flight track area are as follows: The system acquires the real-time latitude and longitude, flight altitude, pitch angle, yaw angle, and roll angle of the aircraft, marks them with a unified timestamp, and calculates the local time. Based on the real-time latitude and longitude of the aircraft, its flight altitude, and local time, the solar vector is calculated and converted from the geographic coordinate system to the direction vector in the aircraft's body coordinate system. Based on the structural parameters of the aircraft and the direction of the solar vector, the solar radiation incident angle and radiation intensity response values ​​of the relevant surfaces of the infrared sensor are calculated, and a solar radiation sensitivity factor matrix is ​​constructed in combination with material properties; Based on the solar radiation sensitivity factor matrix, a three-dimensional grid region is constructed in front of the spacecraft in the current velocity direction. Radiation interference intensity is predicted and rising thermal disturbance potential is identified for each spatial unit, forming a radiation risk mapping field covering the flight track.

3. The real-time meteorological risk assessment method for low-altitude flight safety according to claim 2, characterized in that, The steps for extracting the rising thermal perturbation potential index are as follows: Based on the solar vector direction, the aircraft's heading angle, pitch angle, and roll angle, the radiation interference risk area in the cone-shaped three-dimensional space in front of the aircraft is determined, and this space is divided into scanning units composed of multiple directional vectors. The optical sensing device is controlled to scan each directional vector sequentially at a high frequency, and the infrared irradiance, infrared brightness temperature, thermal noise and image sharpness parameters in each direction are collected. The spatial temperature gradient and the rate of change of radiation intensity are also calculated. A radiation transition surface is constructed based on the radiation intensity gradient, and regions on the transition surface where both the temperature change rate and the vertical gradient exceed a set threshold are identified and extracted as vertical surge regions of thermal perturbation. Calculate the probability of each thermal disturbance vertical surge region intersecting with the aircraft imaging path in the next 3 seconds, and output the optical axis adjustment angle and attitude adjustment suggested parameters.

4. The real-time meteorological risk assessment method for low-altitude flight safety according to claim 3, characterized in that, The steps for pre-scheduling the imaging window are as follows: Based on the spatial coordinates, radiation intensity, thermal gradient direction vector, regional influence radius and predicted duration of disturbance contained in the rising thermal disturbance potential index, a forward projection field of view is constructed and a list of shielded areas is generated. From the remaining directional field of view outside the shielded area, the imaging retention, safety margin, interference change trend and adjustment energy consumption of each directional point are evaluated at preset elevation and azimuth intervals. The directional point with the highest comprehensive score is selected as the target direction of the imaging window. The three-axis angle difference between the current direction of the sensor optical axis and the target direction is calculated and converted into pitch angle, yaw angle and roll angle adjustment amounts. The control parameters are then output to the platform to drive the angle offset control program and complete the imaging window pre-scheduling operation.

5. The real-time meteorological risk assessment method for low-altitude flight safety according to claim 4, characterized in that, After the imaging window is pre-scheduled, the flight attitude fine-tuning control steps are as follows: Based on the optical axis direction after the imaging window is pre-scheduled, the optical axis direction is mapped from the aircraft body coordinate system to the geographic coordinate system through the attitude transformation matrix to obtain the spatial expression of the current imaging direction. Obtain the current flight path direction vector of the aircraft, calculate the angle difference between the imaging direction and the flight path direction in the vertical and horizontal planes, and generate pitch angle deviation and yaw angle deviation; By considering the constraints of the aircraft structure, the control rate boundary, the influence of atmospheric density, and the requirements for interference avoidance, a nonlinear attitude optimization function is constructed to output the optimal attitude target angle. The difference between the target attitude angle and the current attitude is converted into control surface deflection, and the elevator and rudder are controlled to smoothly adjust the aircraft's pitch and yaw angles. After attitude adjustment is completed, the attitude response results are continuously sampled by the inertial measurement unit, the attitude error is calculated, and proportional-integral-derivative feedback control is performed to ensure that the imaging optical axis direction is consistent with the flight trajectory direction.

6. The real-time meteorological risk assessment method for low-altitude flight safety according to claim 5, characterized in that, The steps for generating the signal-to-noise ratio control instruction set are as follows: Image data is acquired using a multi-channel optical observation device, and each frame of the image is compared with the previous few frames to calculate the signal residual of each pixel, generate a residual matrix, and calculate the mean and variance of the residuals. Temperature contrast gradient calculation is performed on each channel image, and the residual matrix and gradient matrix are combined for joint analysis to calculate the local signal-to-noise ratio of each pixel and generate a real-time imaging quality evaluation matrix. Based on the evaluation matrix, the channel with the lowest signal-to-noise ratio is identified and a set of signal-to-noise ratio adjustment instructions is generated, including instructions for exposure time adjustment, frame rate modification, image enhancement, channel switching, and image filter weight adjustment, which are used to optimize imaging quality and feed back to the imaging control path.

7. The real-time meteorological risk assessment method for low-altitude flight safety according to claim 6, characterized in that, According to the signal-to-noise ratio control instruction set, the time-division multiplexing imaging control steps for the spectral and polarization channels are as follows: Based on the real-time imaging quality evaluation matrix, the spectral channel and polarization angle with the optimal signal-to-noise ratio are selected, a complete imaging control cycle is set, and the cycle is divided into multiple sub-time periods, with the spectral channel and polarization angle specified for each sub-time period. A frequency-hopping sampling mechanism is implemented to dynamically adjust the sampling order based on the spatial angle between the solar vector and the current flight attitude, so as to avoid collecting high-risk channel data when there is strong radiation interference. The polarization angle rotation device is controlled to adjust the polarization angle synchronously, and different polarization angles are switched in each sampling cycle to ensure full coverage of imaging requirements for different spectra and polarization states; The micro-shutter array is driven to perform exposure modulation according to the Hadamard sequence, and the exposure duty cycle is dynamically adjusted according to the angle between the solar vector and the sensor optical axis, and the exposure time is adjusted in real time to reduce the interference of solar radiation on the image.

Images