A method for detecting wind shear in civil aviation

Abstract

The invention discloses a wind shear detection method for civil aviation, belonging to the technical field of wind shear detection. The invention divides a target area into a plurality of sub-areas, emits an electromagnetic wave of a fixed frequency to each sub-area, thereby realizing partition detection of the target area, performing spectrum analysis on the echo signal of each sub-area, extracting the deviation frequency and the airflow disturbance index, reflecting the influence of each molecule in the atmosphere on the electromagnetic wave, and then calculating the wind shear risk value according to the distance of each sub-area from an airborne meteorological radar and the deviation frequency, realizing preliminary estimation of the risk by comprehensively considering the distance and the deviation frequency, combining the deviation sequence, the wind shear risk value sequence and the disturbance sequence, adopting a wind shear prediction model to predict the wind shear intensity, and improving the wind shear detection accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of wind shear detection, and in particular to a method for detecting wind shear in civil aviation. Background Art

[0002] In the field of civil aviation, wind shear has always been regarded as an important meteorological factor affecting flight safety. According to statistics, about 30% of aviation accidents in the world are directly related to wind shear, especially during the take-off and landing phases of aircraft, low-altitude wind shear is more likely to cause serious safety hazards. With the rapid development of the aviation transportation industry, there is an urgent need for accurate detection and timely warning of wind shear, which puts higher requirements on existing technologies.

[0003] At present, the wind shear detection technology based on Doppler weather radar mainly relies on radar to obtain reflectivity factor and radial velocity data. After the data is preprocessed (including clutter filtering and quality control), the three-dimensional wind field is reconstructed using the wind field inversion algorithm, and then the features are extracted and the warning output is performed through the wind shear recognition algorithm. However, this technology faces multiple challenges in practical applications: first, there are limitations in data acquisition. The radar can measure the radial wind speed component, but there is velocity ambiguity; second, the uncertainty and large amount of calculation in the wind field inversion process limit its real-time performance. At the same time, the accuracy of the inversion results is highly dependent on the initial field setting. These problems have led to obvious deficiencies in the detection accuracy of the existing wind shear detection technology based on Doppler weather radar. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method for detecting wind shear in civil aviation, which solves the problem of low wind shear detection accuracy in the prior art.

[0005] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a method for detecting wind shear in civil aviation, comprising the following steps:

[0006] S1. Using an airborne weather radar to transmit electromagnetic waves of a fixed frequency to multiple sub-areas in the target area;

[0007] S2, collecting echo signals from each sub-area, performing spectrum analysis on the echo signals, and obtaining spectrum information;

[0008] S3, calculating the deviation frequency and the airflow disturbance index according to the spectrum information and the fixed frequency of the emission;

[0009] S4, taking the deviation frequency of each sub-region as an element to construct a deviation sequence, and taking the airflow disturbance index of each sub-region as an element to construct a disturbance sequence;

[0010] S5. Calculate the wind shear risk value according to the distance of each sub-area from the airborne meteorological radar and the deviation frequency, and construct a wind shear risk value sequence;

[0011] S6. Use the wind shear prediction model to process the deviation sequence, wind shear risk value sequence and disturbance sequence to obtain the wind shear intensity.

[0012] Furthermore, the expression for emitting electromagnetic waves is: , where S i (t) is the electromagnetic wave emitted by the i-th sub-region, rect(t / τ) is the rectangular window function, e is a natural constant, j is an imaginary unit, t is a time variable, τ is a pulse duration, △f is a frequency offset, f0 is the initial carrier frequency, and i is the number of the sub-region.

[0013] Furthermore, S3 includes the following sub-steps:

[0014] S31, extracting the frequency corresponding to the maximum amplitude in the spectrum information corresponding to each echo signal to obtain the main frequency;

[0015] S32, subtracting the main frequency from the transmitted fixed frequency to obtain a deviation frequency;

[0016] S33, removing the main frequency from the spectrum information corresponding to each echo signal to obtain the secondary frequency;

[0017] S34. Calculate an airflow disturbance index according to a secondary frequency corresponding to an echo signal.

[0018] Furthermore, the formula for calculating the airflow disturbance index in S34 is: , where θ is the airflow disturbance index, f n is the nth secondary frequency, f is the fixed frequency of emission, N is the number of secondary frequencies, c is the speed of light, is the average value of the minor frequencies.

[0019] Further, S5 includes the following sub-steps:

[0020] S51, calculating a distance coefficient according to the distance between each sub-area and the airborne weather radar;

[0021] S52, calculating a deviation frequency coefficient according to the deviation frequency of each sub-region;

[0022] S53. Calculate the wind shear risk value according to the distance coefficient and the deviation frequency coefficient;

[0023] S54. Taking the wind shear risk value of a sub-region as an element, constructing a wind shear risk value sequence.

[0024] Furthermore, the formula for calculating the distance coefficient in S51 is: , where μ d is the distance coefficient, d max is the maximum distance, d i is the distance of the ith sub-area from the airborne weather radar, and i is the number of the sub-area;

[0025] The formula for calculating the deviation frequency coefficient in S52 is: , where μ f is the deviation frequency coefficient, f c,max is the maximum deviation frequency, f c,i is the deviation frequency of the ith sub-region;

[0026] The formula for calculating the wind shear risk value in S53 is: , where ε is the wind shear risk value.

[0027] Further, the S6 stroke shear prediction model includes: a first convolutional layer, a second convolutional layer, a third convolutional layer, a first feature fusion layer, a second feature fusion layer, a first fully connected layer, a second fully connected layer and an output layer;

[0028] The input end of the first convolutional layer is used to input the deviation sequence; the input end of the second convolutional layer is used to input the wind shear risk value sequence; the input end of the third convolutional layer is used to input the disturbance sequence;

[0029] A first input end of the first feature fusion layer is connected to an output end of the first convolutional layer, a second input end of the first feature fusion layer is connected to an output end of the second convolutional layer, and an output end of the first feature fusion layer is connected to an input end of the first fully connected layer;

[0030] A first input end of the second feature fusion layer is connected to an output end of the third convolutional layer, a second input end of the second feature fusion layer is connected to an output end of the second convolutional layer, and an output end of the second feature fusion layer is connected to an input end of the second fully connected layer;

[0031] The input end of the output layer is connected to the output end of the first fully connected layer and the output end of the second fully connected layer respectively, and the output end of the output layer serves as the output end of the wind shear prediction model.

[0032] Furthermore, the expression of the first feature fusion layer is: , , , where h 1,m is the mth fusion feature output by the first feature fusion layer, x 1,m is the mth feature output by the first convolutional layer, x 2,m is the mth feature output by the second convolutional layer, w 1,1,m is the first feature fusion layer for x 1,m The first weight term assigned, b1,1,m is the first feature fusion layer for x 1,m The first bias term assigned, w 1,2,m is the first feature fusion layer for x 2,m The second weight term assigned, b 1,2,m is the first feature fusion layer for x 2,m The second bias term assigned, r 1,m is the mth first gating information in the first feature fusion layer, r 2,m is the mth second gating information in the first feature fusion layer, σ is the Sigmoid function, and m is a positive integer;

[0033] The expression of the second feature fusion layer is: , , , where h 2,m is the mth fusion feature output by the second feature fusion layer, x 3,m is the mth feature output by the third convolutional layer, w 2,1,m is the pair x in the second feature fusion layer 2,m The first weight term assigned, b 2,1,m is the pair x in the second feature fusion layer 2,m The first bias term assigned, w 2,2,m is the pair x in the second feature fusion layer 3,m The second weight term assigned, b 2,2,m is the pair x in the second feature fusion layer 3,m The second bias term assigned, g 1,m is the mth first gating information in the second feature fusion layer, g 2,m is the mth second gating information in the second feature fusion layer.

[0034] Furthermore, the expression of the first fully connected layer is: , where s 1,m is the mth feature output by the first fully connected layer, w h1,m is the pair h in the first fully connected layer 1,m The weight term assigned, b h1,m is the pair h in the first fully connected layer 1,m The assigned bias term;

[0035] The expression of the second fully connected layer is: , where s 2,m is the mth feature output by the second fully connected layer, w h2,m is the pair h in the second fully connected layer 2,m The weight term assigned, b h2,m is the pair h in the second fully connected layer 2,m The bias term assigned.

[0036] Furthermore, the expression of the output layer is: , where y is the wind shear intensity, tanh is the hyperbolic tangent activation function, and w s1,m is the pair s in the output layer 1,m The weight term assigned, b s1,m is the pair s in the output layer 1,m The bias term assigned, w s2,m is the pair s in the output layer 2,m The weight term assigned, b s2,m is the pair s in the output layer 2,m The assigned bias term, M1 is the number of features output by the first fully connected layer, and M2 is the number of features output by the second fully connected layer.

[0037] The beneficial effects of the present invention are as follows: the present invention divides the target area into multiple sub-areas, emits an electromagnetic wave of a fixed frequency to each sub-area, thereby realizing partition detection of the target area, performing spectrum analysis on the echo signal of each sub-area, extracting the deviation frequency and the airflow disturbance index, reflecting the influence of each molecule in the atmosphere on the electromagnetic wave, and then calculating the wind shear risk value according to the distance of each sub-area from the airborne meteorological radar and the deviation frequency, comprehensively considering the distance and the deviation frequency, realizing a preliminary estimation of the risk, combining the deviation sequence, the wind shear risk value sequence and the disturbance sequence, adopting the wind shear prediction model to predict the wind shear intensity, and improving the wind shear detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flow chart of a method for detecting wind shear in civil aviation;

[0039] Figure 2 Schematic diagram of the wind shear prediction model. DETAILED DESCRIPTION

[0040] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0041] like Figure 1 As shown, a method for detecting wind shear in civil aviation comprises the following steps:

[0042] S1. Using an airborne weather radar to transmit electromagnetic waves of a fixed frequency to multiple sub-areas in the target area;

[0043] S2, collecting echo signals from each sub-area, performing spectrum analysis on the echo signals, and obtaining spectrum information;

[0044] S3, calculating the deviation frequency and the airflow disturbance index according to the spectrum information and the fixed frequency of the emission;

[0045] S4, taking the deviation frequency of each sub-region as an element to construct a deviation sequence, and taking the airflow disturbance index of each sub-region as an element to construct a disturbance sequence;

[0046] S5. Calculate the wind shear risk value according to the distance of each sub-area from the airborne meteorological radar and the deviation frequency, and construct a wind shear risk value sequence;

[0047] S6. Use the wind shear prediction model to process the deviation sequence, wind shear risk value sequence and disturbance sequence to obtain the wind shear intensity.

[0048] In this embodiment, the expression for transmitting electromagnetic waves is: , where S i (t) is the electromagnetic wave emitted by the i-th sub-region, rect(t / τ) is the rectangular window function, e is a natural constant, j is an imaginary unit, t is a time variable, τ is a pulse duration, △f is a frequency offset, f0 is the initial carrier frequency, and i is the number of the sub-region.

[0049] In this embodiment, the initial carrier frequency f0 is 9.4 GHz, the frequency offset △f is 10 MHz, the fixed frequency of the first sub-area is 9.41 GHz, the fixed frequency of the second sub-area is 9.42 GHz, the fixed frequency of the third sub-area is 9.43 GHz, and the frequency difference between adjacent sub-areas is 10 MHz.

[0050] The present invention realizes frequency grading by allocating different fixed frequencies (f0+iΔf) to each sub-area, thereby improving the anti-interference capability and detection accuracy of the system. Different sub-areas use different frequencies, which can realize simultaneous detection of multiple areas and improve system efficiency.

[0051] The rectangular window function has a value of 1 in the time interval [-τ / 2,τ / 2] and a value of 0 outside this interval.

[0052] In this embodiment, the spectrum analysis method in S2 is Fourier transform, which transforms the time domain signal into the frequency domain.

[0053] In this embodiment, S3 includes the following sub-steps:

[0054] S31, extracting the frequency corresponding to the maximum amplitude in the spectrum information corresponding to each echo signal to obtain the main frequency;

[0055] S32, subtracting the main frequency from the transmitted fixed frequency to obtain a deviation frequency;

[0056] S33, removing the main frequency from the spectrum information corresponding to each echo signal to obtain the secondary frequency;

[0057] S34. Calculate an airflow disturbance index according to a secondary frequency corresponding to an echo signal.

[0058] Since electromagnetic waves of fixed frequency are emitted in the present invention, the present invention extracts the frequency corresponding to the maximum amplitude, which reflects the main signal component of the echo signal. Therefore, the main frequency is selected and subtracted from the fixed frequency of emission to obtain the main influence of the atmospheric environment on the emission frequency, and the secondary frequency is extracted. The secondary frequency contains detailed information of airflow disturbance and reflects the characteristics of atmospheric turbulence and small-scale motion.

[0059] In a sub-region, there is an emitted electromagnetic wave of a fixed frequency and an echo signal, so the processes in S31 to S34 are all for processing the signal of the same sub-region.

[0060] In each sub-area, a deviation frequency, an airflow disturbance index and a wind shear risk value are obtained, so the lengths of the deviation sequence, wind shear risk value sequence and disturbance sequence are consistent.

[0061] In this embodiment, the airflow disturbance index may also be the standard deviation or variance of the spectral component. The standard deviation or variance of a single spectral component may also reflect the airflow disturbance. More preferably, the formula of the airflow disturbance index is: , where θ is the airflow disturbance index, f n is the nth secondary frequency, f is the fixed frequency of emission, N is the number of secondary frequencies, c is the speed of light, is the average value of the minor frequencies.

[0062] The standard deviation of the spectral components is calculated in the present invention, so that the airflow disturbance index can quantify the degree of fluctuation between frequency components. The standard deviation provides a judgment on the central tendency and discreteness of the frequency distribution, which is very critical for the intensity assessment of the disturbance. This formula combines the speed of light and the deviation of the frequency component to effectively quantify the average moving speed of the airflow in the atmosphere. The airflow disturbance index combines the standard deviation and average speed of the spectrum, which not only reflects the fluctuating characteristics of atmospheric turbulence, but also takes into account the overall motion state of the airflow, thereby improving the assessment accuracy of the airflow disturbance.

[0063] The smaller the airflow disturbance index is, the smoother the airflow is.

[0064] In this embodiment, S5 includes the following sub-steps:

[0065] S51, calculating a distance coefficient according to the distance between each sub-area and the airborne weather radar;

[0066] S52, calculating a deviation frequency coefficient according to the deviation frequency of each sub-region;

[0067] S53. Calculate the wind shear risk value according to the distance coefficient and the deviation frequency coefficient;

[0068] S54. Taking the wind shear risk value of a sub-region as an element, constructing a wind shear risk value sequence.

[0069] In this embodiment, the formula for calculating the distance coefficient in S51 is: , where μ d is the distance coefficient, d max is the maximum distance, d i is the distance from the i-th sub-area to the airborne weather radar, and i is the number of the sub-area.

[0070] In this embodiment, the formula for calculating the deviation frequency coefficient in S52 is: , where μ f is the deviation frequency coefficient, f c,max is the maximum deviation frequency, f c,i is the deviation frequency of the ith sub-region.

[0071] In this embodiment, the maximum distance is the farthest distance detected, and the maximum deviation frequency is the maximum degree of influence of the atmosphere on the frequency.

[0072] In this embodiment, the formula for calculating the wind shear risk value in S53 is: , where ε is the wind shear risk value.

[0073] The closer the sub-area is to the airborne meteorological radar, the greater the impact on the fuselage. The larger the distance coefficient and the larger the deviation frequency coefficient, the greater the impact of the atmosphere on the frequency. By combining the frequency impact and distance impact, the wind shear risk value is obtained to assess the current impact on the fuselage.

[0074] In this embodiment, if Figure 2 As shown, the S6 stroke shear prediction model includes: a first convolution layer, a second convolution layer, a third convolution layer, a first feature fusion layer, a second feature fusion layer, a first fully connected layer, a second fully connected layer and an output layer;

[0075] The input end of the first convolutional layer is used to input the deviation sequence; the input end of the second convolutional layer is used to input the wind shear risk value sequence; the input end of the third convolutional layer is used to input the disturbance sequence;

[0076] A first input end of the first feature fusion layer is connected to an output end of the first convolutional layer, a second input end of the first feature fusion layer is connected to an output end of the second convolutional layer, and an output end of the first feature fusion layer is connected to an input end of the first fully connected layer;

[0077] A first input end of the second feature fusion layer is connected to an output end of the third convolutional layer, a second input end of the second feature fusion layer is connected to an output end of the second convolutional layer, and an output end of the second feature fusion layer is connected to an input end of the second fully connected layer;

[0078] The input end of the output layer is connected to the output end of the first fully connected layer and the output end of the second fully connected layer respectively, and the output end of the output layer serves as the output end of the wind shear prediction model.

[0079] The first convolution layer, the second convolution layer and the third convolution layer are one-dimensional convolution layers. The present invention processes different data through three convolution layers, and can extract wind shear-related features from different angles, thereby improving the comprehensiveness and accuracy of the model.

[0080] The first feature fusion layer is used to combine the features generated by the deviation sequence and the features generated by the wind shear risk value sequence. The second feature fusion layer is used to combine the features generated by the wind shear risk value sequence and the features generated by the disturbance sequence. The present invention can gradually extract and integrate information through hierarchical feature fusion. Hierarchical processing helps the model learn more complex feature representations.

[0081] The wind shear risk value sequence reflects the risk size in each sub-area. In the wind shear prediction model, the attention to the elements at important positions in the deviation sequence and the disturbance sequence can be improved. Therefore, the wind shear risk value sequence is combined with the characteristics of the deviation sequence and the disturbance sequence respectively to improve the attention to the characteristics of the key positions in the deviation sequence and the disturbance sequence.

[0082] In this embodiment, the expression of the first feature fusion layer is: , , , where h 1,m is the mth fusion feature output by the first feature fusion layer, x 1,m is the mth feature output by the first convolutional layer, x 2,m is the mth feature output by the second convolutional layer, w 1,1,m is the first feature fusion layer for x 1,m The first weight term assigned, b 1,1,m is the first feature fusion layer for x 1,m The first bias term assigned, w 1,2,m is the first feature fusion layer for x 2,m The second weight term assigned, b 1,2,m is the first feature fusion layer for x 2,m The second bias term assigned, r 1,m is the mth first gating information in the first feature fusion layer, r 2,mis the mth second gating information in the first feature fusion layer, σ is the Sigmoid function, and m is a positive integer.

[0083] The expression of the second feature fusion layer is: , , , where h 2,m is the mth fusion feature output by the second feature fusion layer, x 3,m is the mth feature output by the third convolutional layer, w 2,1,m is the pair x in the second feature fusion layer 2,m The first weight term assigned, b 2,1,m is the pair x in the second feature fusion layer 2,m The first bias term assigned, w 2,2,m is the pair x in the second feature fusion layer 3,m The second weight term assigned, b 2,2,m is the pair x in the second feature fusion layer 3,m The second bias term assigned, g 1,m is the mth first gating information in the second feature fusion layer, g 2,m is the mth second gating information in the second feature fusion layer.

[0084] In the first feature fusion layer and the second feature fusion layer, a gating unit σ is used to process each input feature to obtain gating information. The gating information is used to automatically adjust the feature weight according to the input. It has the function of automatically identifying key features and suppressing irrelevant features, thereby automatically adjusting the ratio of the two features in each feature fusion layer.

[0085] In this embodiment, the expression of the first fully connected layer is: , where s 1,m is the mth feature output by the first fully connected layer, w h1,m is the pair h in the first fully connected layer 1,m The weight term assigned, b h1,m is the pair h in the first fully connected layer 1,m The assigned bias term;

[0086] The expression of the second fully connected layer is: , where s 2,m is the mth feature output by the second fully connected layer, w h2,m is the pair h in the second fully connected layer 2,m The weight term assigned, b h2,m is the pair h in the second fully connected layer 2,m The bias term assigned.

[0087] In this embodiment, the expression of the output layer is: , where y is the wind shear intensity, tanh is the hyperbolic tangent activation function, and w s1,m is the pair s in the output layer 1,m The weight term assigned, b s1,m is the pair s in the output layer 1,m The bias term assigned, w s2,m is the pair s in the output layer 2,m The weight term assigned, b s2,m is the pair s in the output layer 2,m The assigned bias term, M1 is the number of features output by the first fully connected layer, and M2 is the number of features output by the second fully connected layer.

[0088] The output layer in the present invention integrates all the features of the first fully connected layer and the second fully connected layer. By weighted summation of different features, the model can effectively integrate information from different layers, reduce information loss, and improve overall performance.

[0089] In the present invention, the weights and biases in the wind shear prediction model are trained using the gradient descent method.

[0090] In the present invention, wind shear intensity reflects the severity of the change in wind speed.

[0091] The present invention divides a target area into multiple sub-areas, emits an electromagnetic wave of a fixed frequency to each sub-area, thereby realizing partition detection of the target area, performing spectrum analysis on the echo signal of each sub-area, extracting the deviation frequency and the airflow disturbance index, reflecting the influence of each molecule in the atmosphere on the electromagnetic wave, and then calculating the wind shear risk value according to the distance of each sub-area from an airborne meteorological radar and the deviation frequency, realizing a preliminary estimation of the risk by comprehensively considering the distance and the deviation frequency, combining the deviation sequence, the wind shear risk value sequence and the disturbance sequence, adopting a wind shear prediction model to predict the wind shear intensity, and improving the wind shear detection accuracy.

[0092] The present invention does not need to rely on initial field settings, thus avoiding errors caused by inaccurate initial conditions in the prior art. It performs calculations directly based on spectrum analysis and airflow disturbance index, does not require a complex wind field inversion process, and greatly reduces the amount of calculations.

[0093] The present invention directly obtains frequency information by performing spectrum analysis on the echo signal, thus avoiding the velocity ambiguity phenomenon in traditional radial velocity measurement. It adopts the method of fixed-frequency electromagnetic wave emission and frequency deviation calculation to accurately capture the Doppler frequency shift caused by airflow movement, introduces the airflow disturbance index as a supplementary feature, provides additional airflow movement information, and further improves the detection accuracy.

[0094] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for detecting wind shear in civil aviation, characterized in that: The following steps are involved: S1. Using an airborne weather radar to transmit electromagnetic waves of a fixed frequency to multiple sub-areas in the target area; S2, collecting echo signals from each sub-area, performing spectrum analysis on the echo signals, and obtaining spectrum information; S3, calculating the deviation frequency and the airflow disturbance index according to the spectrum information and the fixed frequency of the emission; S4, taking the deviation frequency of each sub-region as an element to construct a deviation sequence, and taking the airflow disturbance index of each sub-region as an element to construct a disturbance sequence; S5. Calculate the wind shear risk value according to the distance of each sub-area from the airborne meteorological radar and the deviation frequency, and construct a wind shear risk value sequence; S6. Using the wind shear prediction model to process the deviation sequence, wind shear risk value sequence and disturbance sequence, obtain the wind shear intensity; The S3 comprises the following sub-steps: S31, extracting the frequency corresponding to the maximum amplitude in the spectrum information corresponding to each echo signal to obtain the main frequency; S32, subtracting the main frequency from the transmitted fixed frequency to obtain a deviation frequency; S33, removing the main frequency from the spectrum information corresponding to each echo signal to obtain the secondary frequency; S34, calculating an airflow disturbance index according to a secondary frequency corresponding to an echo signal; The formula for calculating the airflow disturbance index in S34 is: , where θ is the airflow disturbance index, f n is the nth secondary frequency, f is the fixed frequency of emission, N is the number of secondary frequencies, c is the speed of light, is the average value of the minor frequencies; The S5 comprises the following sub-steps: S51, calculating a distance coefficient according to the distance between each sub-area and the airborne weather radar; S52, calculating a deviation frequency coefficient according to the deviation frequency of each sub-region; S53. Calculate the wind shear risk value according to the distance coefficient and the deviation frequency coefficient; S54, taking the wind shear risk value of a sub-region as an element to construct a wind shear risk value sequence; The formula for calculating the distance coefficient in S51 is: , where μ d is the distance coefficient, d max is the maximum distance, d i is the distance of the ith sub-area from the airborne weather radar, and i is the number of the sub-area; The formula for calculating the deviation frequency coefficient in S52 is: , where μ f is the deviation frequency coefficient, f c,max is the maximum deviation frequency, f c,i is the deviation frequency of the ith sub-region; The formula for calculating the wind shear risk value in S53 is: , where ε is the wind shear risk value; The S6 stroke shear prediction model includes: a first convolution layer, a second convolution layer, a third convolution layer, a first feature fusion layer, a second feature fusion layer, a first fully connected layer, a second fully connected layer and an output layer; The input end of the first convolutional layer is used to input a deviation sequence; the input end of the second convolutional layer is used to input a wind shear risk value sequence; the input end of the third convolutional layer is used to input a disturbance sequence; the first input end of the first feature fusion layer is connected to the output end of the first convolutional layer, its second input end is connected to the output end of the second convolutional layer, and its output end is connected to the input end of the first fully connected layer; the first input end of the second feature fusion layer is connected to the output end of the third convolutional layer, its second input end is connected to the output end of the second convolutional layer, and its output end is connected to the input end of the second fully connected layer; the input end of the output layer is respectively connected to the output end of the first fully connected layer and the output end of the second fully connected layer, and its output end serves as the output end of the wind shear prediction model.

2. The method for detecting wind shear in civil aviation according to claim 1, characterized in that: The expression of the emitted electromagnetic wave is: , where S i (t) is the electromagnetic wave emitted by the i-th sub-region, rect(t / τ) is the rectangular window function, e is a natural constant, j is an imaginary unit, t is a time variable, τ is a pulse duration, △f is a frequency offset, f0 is the initial carrier frequency, and i is the number of the sub-region.

3. The method for detecting wind shear in civil aviation according to claim 1, characterized in that: The expression of the first feature fusion layer is: , , , where h 1,m is the mth fusion feature output by the first feature fusion layer, x 1,m is the mth feature output by the first convolutional layer, x 2,m is the mth feature output by the second convolutional layer, w 1,1,m is the first feature fusion layer for x 1,m The first weight term assigned, b 1,1,m is the first feature fusion layer for x 1,m The first bias term assigned, w 1,2,m is the first feature fusion layer for x 2,m The second weight term assigned, b 1,2,m is the first feature fusion layer for x 2,m The second bias term assigned, r 1,m is the mth first gating information in the first feature fusion layer, r 2,m is the mth second gating information in the first feature fusion layer, σ is the Sigmoid function, and m is a positive integer; The expression of the second feature fusion layer is: , , , where h 2,m is the mth fusion feature output by the second feature fusion layer, x 3,m is the mth feature output by the third convolutional layer, w 2,1,m is the pair x in the second feature fusion layer 2,m The first weight term assigned, b 2,1,m is the pair x in the second feature fusion layer 2,m The first bias term assigned, w 2,2,m is the pair x in the second feature fusion layer 3,m The second weight term assigned, b 2,2,m is the pair x in the second feature fusion layer 3,m The second bias term assigned, g 1,m is the mth first gating information in the second feature fusion layer, g 2,m is the mth second gating information in the second feature fusion layer.

4. The method for detecting wind shear in civil aviation according to claim 3, characterized in that: The expression of the first fully connected layer is: , where s 1,m is the mth feature output by the first fully connected layer, w h1,m is the pair h in the first fully connected layer 1,m The weight term assigned, b h1,m is the pair h in the first fully connected layer 1,m The assigned bias term; The expression of the second fully connected layer is: , where s 2,m is the mth feature output by the second fully connected layer, w h2,m is the pair h in the second fully connected layer 2,m The weight term assigned, b h2,m is the pair h in the second fully connected layer 2,m The bias term assigned.

5. The method for detecting wind shear in civil aviation according to claim 4, characterized in that: The expression of the output layer is: , where y is the wind shear intensity, tanh is the hyperbolic tangent activation function, and w s1,m is the pair s in the output layer 1,m The weight term assigned, b s1,m is the pair s in the output layer 1,m The bias term assigned, w s2,m is the pair s in the output layer 2,m The weight term assigned, b s2,m is the pair s in the output layer 2,m The assigned bias term, M1 is the number of features output by the first fully connected layer, and M2 is the number of features output by the second fully connected layer.

Images