Civil aviation wind shear and visibility early warning method

Abstract

The invention discloses a civil aviation wind shear and visibility early warning method, which comprises the following steps of: acquiring an echo signal by using a high-precision and high-temporal-spatial-resolution coherent laser radar, and simultaneously acquiring civil aviation meteorological conventional observation data; calculating wind field distribution through a wind field inversion algorithm, and calculating visibility distribution through a visibility inversion algorithm; positioning a wind shear region by using a wind shear recognition algorithm, analyzing a visibility space-time evolution trend by combining a space-time deduction algorithm, and generating a visibility profile; inputting the wind shear identification data, the visibility profile data and the meteorological routine observation data into a machine learning model for modeling training; generating a wind shear and visibility early warning scheme over the airport based on the trained prediction model; and generating a wind shear alarm signal and visibility early warning information based on the early warning scheme, and outputting the signals to a civil aviation meteorological early warning system. According to the invention, high-precision wind shear and visibility combined early warning can be realized, and the early warning accuracy under complex weather conditions is improved.

Description

Technical Field

[0001] The invention relates to the technical field of aviation meteorology, and in particular to a civil aviation wind shear and visibility early warning method. Background Art

[0002] In civil aviation operations, wind shear and low visibility are important meteorological factors that affect flight safety, especially during takeoff and landing. Wind shear may cause sudden changes in flight attitude, excessive vertical acceleration and flight trajectory deviation, while low visibility may reduce the pilot's situational awareness and increase the risk of approach and landing. Therefore, accurate warning of wind shear and visibility changes is crucial to ensure aviation safety.

[0003] At present, the detection of wind shear mainly relies on ground-based Doppler weather radar, wind profile radar and the aircraft's own meteorological sensors. However, these methods have limited temporal and spatial resolution, limited detection range and slow response speed, which makes it difficult to meet the rapid warning needs under complex meteorological conditions. In addition, the measurement of visibility usually relies on ground visibility meters and meteorological station observation data. Although these methods can provide more accurate point measurement results, they lack the ability to predict the temporal and spatial evolution trend of visibility. Especially under environmental conditions that are greatly affected by humidity and aerosols, traditional methods often find it difficult to accurately invert a large range of visibility distribution and predict its future change trend. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a civil aviation wind shear and visibility early warning method, comprising the following steps: Use high-precision, high-temporal and high-resolution coherent laser radar to collect echo signals, extract frequency and intensity information from echo signals, and simultaneously obtain routine civil aviation meteorological observation data; Based on the extracted frequency information, the wind field distribution is calculated by the wind field inversion algorithm, and based on the extracted intensity information, the visibility distribution is calculated by the visibility inversion algorithm; Based on the wind field inversion results, the wind shear identification algorithm is used to locate the wind shear area and output the wind shear identification data. At the same time, based on the visibility inversion results, the spatiotemporal evolution trend of visibility is analyzed in combination with the spatiotemporal deduction algorithm to generate the visibility profile and output the visibility profile data; Inputting the wind shear identification data, visibility profile data and conventional meteorological observation data into a machine learning model for modeling training; Generate wind shear and visibility warning plans over the airport based on the trained prediction model; Based on the warning scheme, a wind shear warning signal and visibility warning information are generated and output to the civil aviation meteorological warning system.

[0006] As a preferred solution of the civil aviation wind shear and visibility warning method of the present invention, wherein: the wind field distribution is calculated by a wind field inversion algorithm for the extracted frequency information, comprising the following steps: Preprocess the echo signals collected by high-precision and high-temporal and high-spatial resolution coherent lidar; Using signal processing technology, frequency spectrum analysis is performed on the pre-processed echo signal; Calculate the wind speed distribution along the LiDAR beam direction; Combining data from multiple LiDAR measurement directions, the three-dimensional wind field vector distribution is inverted using the optimal interpolation method; The wind field inversion results are integrated with the conventional civil aviation meteorological observation data to correct the wind field errors.

[0007] As a preferred solution of the civil aviation wind shear and visibility warning method of the present invention, wherein: the visibility distribution is calculated by using a visibility inversion algorithm for the extracted intensity information, comprising the following steps: Preprocessing the echo intensity signal collected by the coherent laser radar; Based on Mie scattering theory, using the atmospheric extinction coefficient and visibility The visibility is calculated by the correlation relationship, and the formula is: , Among them, the extinction coefficient It is obtained by solving the inversion of the lidar equation, specifically using the Klett method or the Fernald method; The visibility is corrected for humidity and compensated for aerosols in combination with routine meteorological observation data.

[0008] As a preferred solution of the civil aviation wind shear and visibility warning method of the present invention, the humidity correction of visibility includes the following steps: Relative humidity is obtained through weather station observation data; The humidity growth factor is used for correction, and the calculation formula is as follows: , in, is the current relative humidity, is the base humidity, is the empirical coefficient; When the humidity is greater than the set value, a nonlinear correction model is introduced, such as: , in, is the correction factor related to the environment; Then, the humidity growth factor Correct the extinction coefficient, the correction formula is: , in, is the original extinction coefficient, measured by lidar; is the extinction coefficient after humidity correction; Finally, calculate the corrected visibility, which is Substitute into Koschmieder formula to calculate visibility.

[0009] As a preferred solution of the civil aviation wind shear and visibility warning method of the present invention, the aerosol compensation for visibility includes the following steps: The aerosol optical depth AOD is obtained through the meteorological station observation data, and the aerosol scattering coefficient is estimated by combining the PM2.5 / PM10 monitoring data: , in, and It is the empirical coefficient, which is based on the regional environmental experience value; is the particle concentration; The aerosol scattering coefficient is corrected using the Angström exponent: , in, is the working wavelength of the laser radar, is the reference wavelength, is the Angström index; Calculate the extinction coefficient after aerosol compensation. When calculating, first calculate the aerosol compensation factor, and then calculate the corrected extinction coefficient based on the aerosol compensation factor; the calculation formulas for the aerosol compensation factor and the corrected extinction coefficient are as follows: , , in, is the experience adjustment coefficient; Substitute into the Koschmieder formula to calculate the visibility after aerosol compensation. The formula is as follows: , in, The visibility is compensated by aerosol. The visibility data after aerosol compensation can avoid misjudgment caused by aerosol interference.

[0010] As a preferred solution of the civil aviation wind shear and visibility warning method of the present invention, the visibility is calculated in combination with the humidity correction and aerosol compensation, and the calculation formula is as follows: , in, In order to improve the reliability of early warning, the visibility data after humidity correction and aerosol compensation are calculated.

[0011] As a preferred solution of the civil aviation wind shear and visibility warning method of the present invention, wherein: the wind field inversion result uses a wind shear identification algorithm to detect the wind shear area and output wind shear identification data, including the following steps: The three-dimensional wind field data is spatially gridded, and vertical profile grids are set along the airport runway and take-off and landing path; Calculate the wind speed gradient and wind direction mutation angle of adjacent grid cells, and set the wind speed gradient and wind direction mutation angle thresholds; If an area meets both the wind speed gradient and wind direction sudden change angle thresholds, it is marked as a potential wind shear area; Verify the time continuity of potential wind shear areas and eliminate transient interference; Combined with the historical wind shear event database, the final wind shear area and its hazard level are determined through spatial pattern matching; The spatial position, wind speed gradient value, wind direction sudden change angle, duration and corresponding danger level of the final wind shear area are integrated to generate wind shear identification data as input of the early warning model.

[0012] As a preferred embodiment of the civil aviation wind shear and visibility warning method described in the present invention, the threshold of the wind speed gradient is set to a wind speed change of not less than 15 knots per 100 meters in the horizontal direction, or a wind speed change of not less than 10 knots per 50 meters in the vertical direction; the threshold of the wind direction mutation angle is set to ≥30°.

[0013] As a preferred solution of the civil aviation wind shear and visibility warning method of the present invention, wherein: based on the visibility inversion result, the visibility spatiotemporal evolution trend is analyzed in combination with the spatiotemporal deduction algorithm to generate a visibility profile, and the visibility profile data is output, including the following steps: The visibility data inverted by LiDAR and the observation data of the weather station are input into the space-time Kriging model to generate the visibility space-time cube of the airport area; Based on the visibility space-time cube, the visibility space-time profile is extracted along the flight take-off and landing path, and the future visibility change trend is predicted by the optical flow method; Perform morphological expansion processing on low visibility areas to identify continuous dangerous areas; Combining the spatiotemporal changes of humidity and aerosol data, the prediction results of the optical flow method are corrected to generate a visibility evolution probability map; Based on the visibility evolution probability map, the visibility profile on the flight path is extracted, and the mutation points and continuous low visibility sections are marked; The spatial position, forecast time, visibility value distribution, mutation point position and low-value section information of the visibility profile are integrated to generate visibility profile data as one of the input data sources of the early warning model.

[0014] As a preferred solution of the civil aviation wind shear and visibility warning method of the present invention, the extraction of visibility time-space profile includes the following steps: Set spatial sampling lines according to airport runways and standard approach and departure procedures to ensure coverage of critical flight phases; Based on the generated visibility space-time cube, three-dimensional linear interpolation is performed along the path to obtain the continuous visibility space-time distribution; With horizontal distance as the abscissa, height as the ordinate and time series as the third dimension, a spatiotemporal profile of visibility is constructed to characterize the vertical distribution and temporal evolution characteristics of visibility along the route.

[0015] Beneficial effects of the present invention: 1. The present invention accurately locates the wind shear area through the wind field inversion algorithm and the wind shear identification algorithm, and uses the visibility inversion algorithm and the visibility spatiotemporal deduction algorithm in combination with humidity correction and aerosol compensation to dynamically predict the visibility evolution trend of the airport's key routes, thereby achieving high-precision wind shear and visibility joint warning and improving the warning accuracy under complex weather conditions; 2. The present invention uses the optical flow method to predict the future visibility change trend and combines the humidity and aerosol influencing factors for dynamic correction to achieve short-term visibility warning, and accurately marks the low visibility mutation points and continuous low visibility segments through the low-value area morphological expansion processing, thereby providing a more accurate visibility evolution prediction for aircraft approach and landing. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them: Figure 1 A flowchart of a civil aviation wind shear and visibility warning method of the present invention; Figure 2 A flowchart of a method for early warning of wind shear and visibility in civil aviation according to the present invention for calculating wind field distribution through a wind field inversion algorithm; Figure 3 The present invention is a flowchart of a method for early warning of wind shear and visibility in civil aviation, which uses a visibility inversion algorithm to calculate visibility distribution. DETAILED DESCRIPTION

[0017] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0018] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0019] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0020] Secondly, the present invention is described in detail with reference to the schematic diagram. When describing the embodiments of the present invention in detail, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0021] Example 1

[0022] Reference Figure 1-3 , as an embodiment of the present invention, provides a civil aviation wind shear and visibility warning method, comprising the following steps:

[0023] S1: Use high-precision and high-temporal and high-resolution coherent lidar to collect echo signals and extract frequency and intensity information from the echo signals. At the same time, obtain routine civil aviation meteorological observation data to provide more comprehensive and accurate information for subsequent meteorological analysis and early warning.

[0024] Specifically, routine civil aviation meteorological observation data, including ground wind, visibility, precipitation, temperature, humidity and pressure information, is processed by partition noise filtering to improve data quality.

[0025] It should be noted that high-precision and high-temporal and spatial resolution coherent laser radar is an advanced remote sensing detection equipment. High precision means that the error of its measurement results is small and it can provide accurate meteorological parameter information; high temporal and spatial resolution means that the radar has high resolution in both time and space dimensions, and can make multiple observations of a small area in a short time, thereby capturing the rapid changes and subtle differences in meteorological elements. Coherent laser radar emits a laser beam and receives the echo signal reflected by the target, and uses coherent detection technology to extract the frequency and intensity information in the echo signal. It belongs to the existing technology and will not be described in detail here.

[0026] S2: Based on the extracted frequency information, the wind field distribution is calculated by the wind field inversion algorithm, and based on the extracted intensity information, the visibility distribution is calculated by the visibility inversion algorithm.

[0027] S21: Calculate the wind field distribution using a wind field inversion algorithm based on the extracted frequency information, including the following steps:

[0028] S211: De-noise, filter and normalize the echo signals collected by high-precision and high-temporal and high-resolution coherent lidar to improve signal quality and provide a reliable data basis for subsequent analysis.

[0029] S212: Use signal processing techniques such as fast Fourier transform or wavelet transform to perform frequency spectrum analysis on the pre-processed echo signal and accurately extract the Doppler frequency shift information in the signal. This frequency shift information reflects the motion state of particles in the atmosphere and is the key basis for subsequent wind speed calculation.

[0030] S213: Based on Doppler shift formula Calculate the wind speed distribution along the lidar beam direction, where is the laser wavelength, is the Doppler frequency shift. Through this formula, the frequency shift information can be converted into actual wind speed values ​​to obtain the wind speed distribution along the beam direction.

[0031] S214: Combining data from multiple laser radar measurement directions, using an optimal interpolation method to invert the three-dimensional wind field vector distribution.

[0032] It should be noted that the optimal interpolation method is an objective analysis method based on the least squares principle. It is an existing technology and has been widely used in the assimilation and inversion of meteorological data. Through this method, the three-dimensional wind field vector distribution can be accurately inverted. The three-dimensional wind field vector distribution contains spatial grid data of wind speed and direction for subsequent wind shear identification.

[0033] S215: Integrate the wind field inversion results with the conventional civil aviation meteorological observation data to correct the wind field error and improve the accuracy of wind field distribution. The wind field error correction can be implemented using existing technology, which will not be described in detail here.

[0034] S22: Calculating visibility distribution using a visibility inversion algorithm for the extracted intensity information, including the following steps:

[0035] S221: Preprocess the echo intensity signal collected by the coherent laser radar.

[0036] During preprocessing, first, a noise filtering algorithm is used to accurately remove background noise from the signal, effectively reducing the interference of noise on subsequent calculations. Then, a normalization processing method is used to adjust the echo intensity signal to a unified dimension and range to ensure that signals at different times and locations are comparable, laying the foundation for subsequent accurate calculations.

[0037] S222: Based on Mie scattering theory, using the atmospheric extinction coefficient and visibility The visibility is calculated by the correlation relationship, and the formula is: , Among them, the extinction coefficient It is obtained by solving the lidar equation inversion, specifically using the Klett method or the Fernald method.

[0038] S223: Combining conventional meteorological observation data (such as humidity and aerosol concentration) to perform humidity correction and aerosol compensation on visibility can improve the accuracy of visibility measurement, help to more accurately predict weather changes, and help pilots and air traffic controllers better understand the weather conditions around the airport and make safer flight decisions.

[0039] Specifically, the visibility is corrected for humidity and compensated for aerosol, including the following steps: When making humidity corrections: Relative humidity (RH) is obtained through weather station observation data; The humidity growth factor is used for correction, and the calculation formula is as follows: , in, is the current relative humidity (%), The reference humidity is usually 50%. It is an empirical coefficient, generally ranging from 0.6 to 1.2, depending on the type of ambient aerosol.

[0040] When the humidity is greater than the set value, the set value is set to 80%, that is, when When >80%, a nonlinear correction model is introduced, such as: , in, It is an environmental correction factor, usually ranging from 0.1 to 0.3.

[0041] Then, the humidity growth factor Corrected extinction coefficient: , in, is the original extinction coefficient, measured by lidar; is the extinction coefficient after humidity correction.

[0042] Finally, calculate the corrected visibility, which is Substitute the Koschmieder formula to calculate visibility: , When performing aerosol compensation: The aerosol optical depth (AOD) is obtained through meteorological station observation data, and the aerosol scattering coefficient is estimated in combination with PM2.5 / PM10 monitoring data: , in: and It is the empirical coefficient, which is based on the regional environmental experience value; is the particle concentration ( ).

[0043] The aerosol scattering coefficient is corrected using the Angström exponent: , in, The operating wavelength of the laser radar (such as 355nm, 532nm); is the reference wavelength, usually 550nm; is the Angström index, which is generally 1.0-1.5.

[0044] When calculating the extinction coefficient after aerosol compensation, the aerosol compensation factor is first calculated, and then the corrected extinction coefficient is calculated based on the aerosol compensation factor. The calculation formulas for the aerosol compensation factor and the corrected extinction coefficient are as follows: , , in, It is an empirical adjustment coefficient, usually 0.1-0.3.

[0045] Substitute into the Koschmieder formula to calculate the visibility after aerosol compensation: , in, The visibility is compensated by aerosol. The visibility data after aerosol compensation can avoid misjudgment caused by aerosol interference.

[0046] Finally, the visibility is calculated by combining humidity correction and aerosol compensation. The calculation formula is as follows: , in, In order to improve the reliability of early warning, the visibility data after humidity correction and aerosol compensation are calculated.

[0047] The above humidity correction + aerosol compensation method improves the visibility inversion accuracy under extreme weather conditions, thereby ensuring the reliability of airport low visibility warnings.

[0048] S3: Based on the wind field inversion results, the wind shear identification algorithm is used to locate the wind shear area and output the wind shear identification data. At the same time, based on the visibility inversion results, the spatiotemporal evolution trend of visibility is analyzed in combination with the spatiotemporal deduction algorithm to generate the visibility profile and output the visibility profile data.

[0049] Specifically, based on the wind field inversion result, detecting the wind shear area using the wind shear identification algorithm includes the following steps:

[0050] S311: Divide the three-dimensional wind field data into spatial grids and set vertical profile grids along the airport runway and take-off and landing path (horizontal resolution ≤ 100m, vertical resolution ≤ 50m).

[0051] S312: Calculate the wind speed gradient and wind direction change angle of adjacent grid cells.

[0052] Specifically, the wind speed gradient threshold is set to ≥15 knots / 100m (horizontally) or ≥10 knots / 50m (vertically); The wind direction sudden change angle threshold is set to ≥30°.

[0053] When the wind speed gradient and wind direction mutation in a certain area reach the preset threshold at the same time, that is, the horizontal wind speed change within 100 meters is not less than 15 knots, or the vertical wind speed change within 50 meters is not less than 10 knots, and the wind direction change angle between adjacent grid cells is not less than 30°, the area will be marked as a potential wind shear area.

[0054] For the marked potential wind shear areas, further time continuity verification is required. Specifically, it is necessary to confirm whether the wind shear phenomenon in the area persists for at least 2 minutes, so as to eliminate misjudgments caused by transient interference and ensure the accuracy of the identification results.

[0055] After confirming the time persistence of the potential wind shear area, it is necessary to combine the historical wind shear event database and use spatial pattern matching technology to conduct a detailed analysis of the potential area. By comparing historical data with current observation data, the specific scope of the wind shear area is finally determined, and the danger level is scientifically divided into low, medium and high levels according to its impact and potential risks, providing a strong basis for subsequent risk prevention and response measures.

[0056] The spatial position of the final wind shear area, wind speed gradient value, wind direction mutation angle, duration and corresponding danger level are integrated to generate wind shear identification data. The wind shear identification data is used to describe the spatial distribution characteristics and intensity change characteristics of wind shear over the airport, and serves as the input of the early warning model.

[0057] Furthermore, based on the visibility inversion results, the spatiotemporal evolution trend of visibility is analyzed and the visibility profile is generated by combining the spatiotemporal deduction algorithm, including the following steps:

[0058] S321: Input the visibility data of discrete points (visibility data inverted by lidar and observation data of meteorological stations) into the space-time Kriging model to generate the visibility space-time cube of the airport area (time resolution ≤ 5 minutes, spatial resolution ≤ 200m).

[0059] It should be noted that the space-time Kriging model is an extension of the traditional Kriging method in the space-time dimension. Its core is to model the coupling of spatial correlation and temporal correlation through the space-time covariance function. This method can simultaneously consider the data correlation in time and space, and perform linear unbiased optimal estimation of unsampled points. Generating the visibility space-time cube of the airport area through the space-time Kriging model can help improve the operating efficiency and safety of the airport and reduce the risk of flight delays and accidents caused by low visibility weather. It belongs to the existing technology and will not be described in detail here.

[0060] S322: Based on the visibility space-time cube, the visibility space-time profile is extracted along the flight take-off and landing path, and the visibility change trend in the next 15-30 minutes is predicted by the optical flow method.

[0061] Specifically, extracting the visibility spatiotemporal profile includes the following steps: Set spatial sampling lines according to airport runways and standard approach and departure procedures (such as ILS approach paths) to ensure coverage of key flight phases, such as takeoff, approach, and landing; Based on the generated visibility space-time cube, three-dimensional linear interpolation is performed along the path to obtain the continuous visibility space-time distribution; With horizontal distance (along the runway axis) as the abscissa, altitude (AGL) as the ordinate and time series as the third dimension, a visibility spatiotemporal profile is constructed to characterize the vertical distribution and temporal evolution characteristics of visibility along the route.

[0062] S323: Perform morphological expansion processing on low visibility areas (≤1000m) to identify continuous dangerous areas. The morphological expansion processing method adopts the well-known technology in the field, and can refer to Section 4.3 (Identification of Dangerous Weather Areas) of ICAO Doc 9837 "Manual of Aeronautical Meteorological Information" and Chapter 16 (Spatial Analysis of Meteorological Data) of WMO-No.1200 "Meteorological Instruments and Observation Methods".

[0063] S324: Combine the spatiotemporal changes of humidity and aerosol data, modify the prediction results of the optical flow method, and generate a visibility evolution probability map. The visibility evolution probability map is generated using ensemble forecasting technology, reference: Zhou et al. "Atmospheric Environment" 2021 (aerosol-visibility relationship model).

[0064] S325: Based on the visibility evolution probability map, the visibility profile on the flight path is extracted, and the mutation points (areas with probability spatial gradient ≥ 0.2% / m) and continuous low visibility segments (continuous segments with probability > 70% and duration ≥ 10 minutes) are marked. The profile extraction and marking methods adopt the standard processing procedures in aviation meteorological services (see ICAO Annex 3 Appendix C).

[0065] S326: Integrate the spatial position, prediction time, visibility value distribution, mutation point position and low-value section information of the visibility profile to generate visibility profile data. The visibility profile data is used to characterize the visibility change trend along the route and serves as one of the input data sources of the early warning model.

[0066] S4: Input wind shear identification data, visibility profile data and conventional meteorological observation data into the machine learning model for modeling training.

[0067] Specifically, wind shear identification data, visibility profile data and conventional meteorological observation data are input into the machine learning model for modeling training, including the following steps:

[0068] S41: Clean, denoise, normalize and align wind shear identification data, visibility profile data and conventional meteorological observation data to ensure consistent data quality and eliminate outliers, providing reliable input for subsequent feature extraction and training.

[0069] S42: Extract key features (core variables that affect wind shear and visibility changes) from preprocessed data, such as wind shear gradient, visibility spatiotemporal change rate, meteorological element correlation parameters, etc., and perform dimensionality reduction and feature selection. Dimensionality reduction uses principal component analysis to reduce data dimensions and retain main information. Feature selection uses recursive feature elimination or feature importance evaluation based on tree models to select the most predictive features and improve model training efficiency.

[0070] S43: Divide the data into training set, validation set and test set in the ratio of 7:1.5:1.5, and annotate the categories or severity of wind shear and low visibility events based on the historical meteorological event database to provide reliable labels for supervised learning.

[0071] S44: Select the generalized regression neural network (GRNN) as the basic model to build the prediction model, and define the loss function, optimization method and hyperparameters.

[0072] The loss function definition is to select the mean square error or cross entropy loss function according to the task requirements, and the difference between the predicted value and the true value. The optimization method uses the Adam optimizer, combined with the learning rate decay strategy to accelerate the convergence of the model. The hyperparameter setting can determine the key hyperparameters such as the smoothing factor of GRNN through network search or random search.

[0073] S45: Use the training set for multiple rounds of iterative training, adjust the model parameters through the gradient descent algorithm, and combine K-fold cross validation to evaluate the performance of the model on different data subsets to prevent overfitting. Set up performance monitoring on the validation set and stop training in advance when the performance no longer improves to avoid overfitting.

[0074] S46: Calculate the prediction error, confusion matrix, AUC-ROC and other indicators on the test set to comprehensively evaluate the model performance. According to the evaluation results, adjust the model structure (such as adding hidden layers, changing activation functions) or hyperparameters (such as learning rate, regularization coefficient) to optimize the model effect.

[0075] S47: Combine multiple GRNN models for ensemble learning, use weighted average or voting mechanism to improve prediction stability and accuracy, and obtain the final GRNN neural network prediction model. The final prediction model is deployed to the early warning system interface.

[0076] Through the above steps, a more accurate and efficient wind shear and visibility joint warning model can be constructed.

[0077] S5: Generate wind shear and visibility warning plans over the airport based on the trained prediction model.

[0078] S6: Generate a wind shear warning signal and visibility warning information based on the warning scheme, and output them to the civil aviation meteorological warning system.

[0079] In summary, the present invention is based on the fusion of multi-source data such as coherent laser radar and civil aviation meteorological observation data, and combined with machine learning models, the wind shear area is accurately located through the wind field inversion algorithm and the wind shear identification algorithm, and the visibility inversion algorithm and the visibility spatiotemporal deduction algorithm are combined with humidity correction and aerosol compensation to dynamically predict the visibility evolution trend of the airport's key routes, thereby achieving high-precision wind shear and visibility joint warnings, and improving the accuracy of warnings under complex weather conditions. The present invention generates a visibility spatiotemporal cube of the airport area through a spatiotemporal Kriging model to achieve continuous visibility change modeling over time and space, uses the optical flow method to predict future visibility change trends and combines humidity and aerosol influencing factors for dynamic correction, and achieves minute-level short-term visibility warnings, and accurately marks low visibility mutation points and continuous low visibility segments through low-value area morphological expansion processing, thereby providing more accurate visibility evolution predictions for aircraft approach and landing. The present invention constructs a humidity correction method based on humidity growth factor and nonlinear correction model to solve the problem of low visibility calculation under high humidity conditions. At the same time, an aerosol compensation method combining Angström index correction with aerosol optical thickness (AOD) parameterized model is used to reduce visibility errors caused by polluted weather. The dynamic compensation factors of humidity and aerosol are integrated to construct a unified visibility correction formula, thereby realizing adaptive correction under different environmental conditions, improving the visibility inversion accuracy under extreme weather conditions, and ensuring the reliability of low visibility warning at airports.

[0080] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A civil aviation wind shear and visibility early warning method, characterized in that: The following steps are involved: Use high-precision, high-temporal and high-resolution coherent laser radar to collect echo signals, extract frequency and intensity information from echo signals, and simultaneously obtain routine civil aviation meteorological observation data; Based on the extracted frequency information, the wind field distribution is calculated by the wind field inversion algorithm, and based on the extracted intensity information, the visibility distribution is calculated by the visibility inversion algorithm; Based on the wind field inversion results, the wind shear identification algorithm is used to locate the wind shear area and output the wind shear identification data. At the same time, based on the visibility inversion results, the spatiotemporal evolution trend of visibility is analyzed in combination with the spatiotemporal deduction algorithm to generate the visibility profile and output the visibility profile data; Inputting the wind shear identification data, visibility profile data and conventional meteorological observation data into a machine learning model for modeling training; Generate wind shear and visibility warning plans over the airport based on the trained prediction model; Based on the warning scheme, a wind shear warning signal and visibility warning information are generated and output to the civil aviation meteorological warning system.

2. The civil aviation wind shear and visibility early warning method according to claim 1, characterized in that: The method of calculating the wind field distribution by using a wind field inversion algorithm for the extracted frequency information comprises the following steps: Preprocess the echo signals collected by high-precision and high-temporal and high-spatial resolution coherent lidar; Using signal processing technology, frequency spectrum analysis is performed on the pre-processed echo signal; Calculate the wind speed distribution along the LiDAR beam direction; Combining data from multiple LiDAR measurement directions, the three-dimensional wind field vector distribution is inverted using the optimal interpolation method; The wind field inversion results are integrated with the conventional civil aviation meteorological observation data to correct the wind field errors.

3. The civil aviation wind shear and visibility early warning method according to claim 1, characterized in that: The method of calculating the visibility distribution by using a visibility inversion algorithm for the extracted intensity information comprises the following steps: Preprocessing the echo intensity signal collected by the coherent laser radar; Based on Mie scattering theory, using the atmospheric extinction coefficient and visibility The visibility is calculated by the correlation relationship, and the formula is: ， in, is a fixed constant that reflects the human eye's ability to perceive contrast. is the extinction coefficient, which is obtained by solving the lidar equation inversion, specifically using the Klett method or the Fernald method; The visibility is corrected for humidity and compensated for aerosols in combination with routine meteorological observation data.

4. The civil aviation wind shear and visibility early warning method according to claim 3, characterized in that: The humidity correction for visibility comprises the following steps: Relative humidity is obtained through weather station observation data; The humidity growth factor is used for correction, and the calculation formula is as follows: ， in, is the current relative humidity, is the base humidity, is the empirical coefficient; When the humidity is greater than the set value, a nonlinear correction model is introduced; ， in, is the correction factor related to the environment; Then, the humidity growth factor Correct the extinction coefficient, the correction formula is: ， in, is the original extinction coefficient, measured by lidar; is the extinction coefficient after humidity correction; Finally, calculate the corrected visibility, which is Substitute into Koschmieder formula to calculate visibility.

5. The civil aviation wind shear and visibility early warning method according to claim 4, characterized in that: The aerosol compensation for visibility comprises the following steps: The aerosol optical depth AOD is obtained through the meteorological station observation data, and the aerosol scattering coefficient is estimated by combining the PM2.5 / PM10 monitoring data: ， in, and is the empirical coefficient, based on the regional environmental experience value; is the particle concentration; The aerosol scattering coefficient is corrected using the Angström exponent: ， in, is the working wavelength of the laser radar, is the reference wavelength, is the Angström index; Calculate the extinction coefficient after aerosol compensation. When calculating, first calculate the aerosol compensation factor, and then calculate the corrected extinction coefficient based on the aerosol compensation factor; the calculation formulas for the aerosol compensation factor and the corrected extinction coefficient are as follows: ， ， in, is the experience adjustment coefficient; Substitute into the Koschmieder formula to calculate the visibility after aerosol compensation. The formula is as follows: ， in, The visibility is compensated by aerosol. The visibility data after aerosol compensation can avoid misjudgment caused by aerosol interference.

6. The civil aviation wind shear and visibility early warning method according to claim 5, characterized in that: The visibility is calculated by combining the humidity correction and aerosol compensation, and the calculation formula is as follows: ， in, In order to improve the reliability of early warning, the visibility data after humidity correction and aerosol compensation are calculated.

7. The civil aviation wind shear and visibility early warning method according to claim 2, characterized in that: Based on the wind field inversion result, a wind shear area is detected using a wind shear identification algorithm, and wind shear identification data is output, including the following steps: The three-dimensional wind field data is spatially gridded, and vertical profile grids are set along the airport runway and take-off and landing path; Calculate the wind speed gradient and wind direction mutation angle of adjacent grid cells, and set the wind speed gradient and wind direction mutation angle thresholds; If an area meets both the wind speed gradient and wind direction sudden change angle thresholds, it is marked as a potential wind shear area; Verify the time continuity of potential wind shear areas and eliminate transient interference; Combined with the historical wind shear event database, the final wind shear area and its hazard level are determined through spatial pattern matching; The spatial position, wind speed gradient value, wind direction sudden change angle, duration and corresponding danger level of the final wind shear area are integrated to generate wind shear identification data as input of the early warning model.

8. The civil aviation wind shear and visibility early warning method according to claim 7, characterized in that: The threshold of the wind speed gradient is set to a wind speed change of not less than 15 knots per 100 meters in the horizontal direction, or a wind speed change of not less than 10 knots per 50 meters in the vertical direction; the threshold of the wind direction mutation angle is set to ≥30°.

9. The civil aviation wind shear and visibility early warning method according to claim 3, characterized in that: Based on the visibility inversion result, the visibility spatiotemporal evolution trend is analyzed in combination with the spatiotemporal deduction algorithm to generate a visibility profile, and the visibility profile data is output, including the following steps: The visibility data inverted by LiDAR and the observation data of the weather station are input into the space-time Kriging model to generate the space-time cube of visibility in the airport area; Based on the visibility space-time cube, the visibility space-time profile is extracted along the flight take-off and landing path, and the future visibility change trend is predicted by the optical flow method; Perform morphological expansion processing on low visibility areas to identify continuous dangerous areas; Combining the spatiotemporal changes of humidity and aerosol data, the prediction results of the optical flow method are corrected to generate a visibility evolution probability map; Based on the visibility evolution probability map, the visibility profile on the flight path is extracted, and the mutation points and continuous low visibility sections are marked; The spatial position, forecast time, visibility value distribution, mutation point position and low-value section information of the visibility profile are integrated to generate visibility profile data as one of the input data sources of the early warning model.

10. The civil aviation wind shear and visibility early warning method according to claim 9, characterized in that: The extraction of visibility spatiotemporal profile comprises the following steps: Set spatial sampling lines according to airport runways and standard approach and departure procedures to ensure coverage of critical flight phases; Based on the generated visibility space-time cube, three-dimensional linear interpolation is performed along the path to obtain the continuous visibility space-time distribution; With horizontal distance as the abscissa, height as the ordinate and time series as the third dimension, a spatiotemporal profile of visibility is constructed to characterize the vertical distribution and temporal evolution characteristics of visibility along the route.

Images