A New Method for Extracting Regional Atmospheric Waveguides Based on AIS Signals

Abstract

This specification introduces a new method for extracting regional atmospheric ducts based on AIS signals, including: S1, processing and screening of AIS data; S2, constructing a dataset of the modified refractive index profile of the atmospheric environment waveguide; S4, constructing a dataset of the AIS signal propagation environment and classifying the dataset; S5, establishing a generative adversarial network to obtain a relatively accurate simulation result of the atmospheric environment; S6, verifying the accuracy of the simulation result. The present invention constructs the situation of suspended waveguides within a regional scope by utilizing the feature that AIS signals can be used as an indicator factor for suspended waveguides and surface waveguides with a basis, avoiding the interference of other types of waveguides and enhancing the temporal and spatial resolution of waveguide prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of radio wave propagation, and particularly relates to a new method for extracting regional atmospheric ducts based on AIS signals. Background Art

[0002] An atmospheric duct is a special atmospheric layer structure in which the atmospheric refractive index rapidly decreases with the increase of altitude within a certain altitude range. Its special "pipe-like" structure causes the radio wave to refract during propagation, so that electromagnetic waves in certain frequency bands are trapped in the duct and can propagate with less energy attenuation in the propagation direction, enabling the electromagnetic wave to reach distances that are usually difficult to reach. The characteristics of the atmospheric duct make it have important research value. It is a marine environmental factor that must be considered by commanders when formulating combat plans in the military, and it is also an important influencing factor for enhancing or reducing the performance of civilian communication equipment.

[0003] Currently, the existing research on the inversion of marine atmospheric ducts mainly includes the following two directions: contact detection and remote sensing detection. Contact detection is to use the meteorological parameters directly measured by meteorological sensors to construct a profile to judge whether there is an atmospheric duct structure, or substitute the atmospheric meteorological parameters at a certain altitude into a mathematical model to judge the existence of the atmospheric duct; remote sensing detection is to use technical equipment such as meteorological satellites, radars, and microwave radiometers to invert the atmospheric duct.

[0004] Although the above-mentioned atmospheric duct inversion methods have been mature, there are still many problems, such as: the real-time monitoring ability and the wide-area coverage ability still cannot meet the actual requirements. Most of them are only applied to scientific research and have not yet formed the ability of large-scale commercial operation. They cannot meet the guarantee requirements of the military and civilian for atmospheric ducts, seriously restricting the use ability of China's marine rights and interests. Therefore, it is urgent to study new atmospheric duct detection technologies that can meet the requirements.

[0005] The Automatic Identification System (AIS) for ships is a new type of navigation assistance device. It can automatically provide relevant information on the data generated by the satellite navigation system and the basic ship data to other ships, coastal stations, and aircraft equipped with corresponding devices using Very High Frequency (VHF). Currently, nearly 300 AIS base stations have been built along the coast of China, achieving full regional coverage of China's sea areas. The number of ships receiving AIS signals per day is nearly 70,000, of which nearly 40,000 are ships on coastal routes. The VHF (Very High Frequency) signals emitted by the AIS system have a low frequency and are not affected by evaporation ducts, but can be affected by elevated ducts and surface ducts with a basic layer, resulting in over-the-horizon phenomena. Therefore, elevated ducts and surface ducts with a basic layer have a strong correlation with AIS signals, which can further improve the theoretical system of atmospheric duct detection and provide technical support for the next step of building an atmospheric duct monitoring network using domestic AIS base stations. Summary of the Invention

[0006] The present invention provides a new method for extracting regional atmospheric ducts based on AIS signals to solve the core technical problem of realizing atmospheric duct monitoring through AIS data, forming an internal quantitative relationship between shore-based AIS data and atmospheric duct characteristic quantity information, and providing technical support for the next step of building an atmospheric duct monitoring network using domestic AIS base stations.

[0007] To achieve the above object, the present invention is implemented by the following technical solutions:

[0008] A new method for extracting regional atmospheric ducts based on AIS signals, the steps include:

[0009] S1, Obtain AIS record data and perform screening and classification processing on the data;

[0010] S2, Obtain atmospheric environment data and construct a corrected atmospheric refractive index data set with a certain spatial resolution;

[0011] S3, Construct a data set of atmospheric corrected refractive index profiles associated with AIS signals, generate an AIS signal propagation environment database using AIS position information and atmospheric corrected refractive index profiles, screen the data, and divide the data set into regions;

[0012] S4, Use a generative adversarial network to establish a generator and a discriminator. The generator is used to construct a profile that is closest to the corrected atmospheric refractive index profile in the existing AIS signal propagation environment data set in this region, and the discriminator is responsible for judging whether the fitting result is accurate;

[0013] S5. Verification of the Fitted Profile Results. According to the position information recorded by AIS in the dataset, the parabolic equation is used to calculate the power loss of radio wave propagation by using the new profile and the original profile respectively. The similarity of the two power loss curves is compared by the Euclidean distance method. If the Euclidean distance calculation results of the loss curves at the positions recorded by each AIS in the area are all less than 0.5, then this new profile is regarded as the atmospheric duct result of the corresponding area.

[0014] Further, in S1, the AIS cipher is decoded to obtain the recorded ship data. The specific method includes:

[0015] The first step is to process the AIS cipher to obtain the recorded Coordinated Universal Time (UTC) and the longitude and latitude of the ship's position.

[0016] The second step is to calculate the distance between the ship and the base station according to the longitude and latitude. Since the normal propagation distance of the AIS signal is 80 km, if the recorded ship-base distance is greater than 80 km, it must be affected by the atmospheric duct to increase the signal propagation distance. The data with a distance greater than 80 km can be marked as the data affected by the atmospheric duct in the signal propagation, and the data with a distance less than 80 km is marked as the data not affected by the atmospheric duct in the signal propagation process.

[0017]

[0018] Calculate the distance L between the ship and the AIS base station according to the longitude and latitude information i As shown in formula (1), where (W i , W j ) and (J i , J j ) are the longitude and latitude of the propagation position recorded by AIS and the receiving base station, and R 地 earth is the radius of the earth, and the average value can be taken as 6371 km.

[0019] Further, in S2, the ERA5 reanalysis meteorological data of the European Centre for Medium-Range Weather Forecasts (ECMWF) is used to model the profile hour by hour to construct the atmospheric environment. The specific method includes:

[0020] The first step: Read the ERA5 data. Use the hourly ERA5 meteorological reanalysis data released by the European Centre for Medium-Range Weather Forecasts. The horizontal resolution of the meteorological data used is 0.25°. Calculate the corrected atmospheric refractive index profile by using the temperature, pressure, and water vapor pressure data of the lowest 12 atmospheric pressure layers of the ERA5 reanalysis data, and identify and record the characteristic parameters of the atmospheric duct through interpolation. The formula for the corrected atmospheric refractive index is as follows:

[0021]

[0022] In the formula, M is the atmospheric correction refractive index, which is a dimensionless quantity and is usually denoted as M unit; T is the absolute atmospheric temperature (K); P is the atmospheric pressure (hPa); e is the water vapor pressure (hPa); z is the altitude (m);

[0023] Step 2: Classify the atmospheric duct conditions. Calculate according to the calculation result M. When then it is considered that M n-1 atmospheric duct phenomenon occurs at the height of when, M n-1 the atmospheric duct phenomenon ends at the height of

[0024] M n-1 and M n are two atmospheric refractive indices at adjacent heights in the calculation result of the vertical distribution of the corrected atmospheric refraction index. M n-1 is the atmospheric refractive index at the higher point, M n is the atmospheric refractive index at the lower point, and Δz is the height difference between the above adjacent heights.

[0025] Therefore, M(z) increases with the decrease of height in a certain height range, which indicates that there is a duct layer at this height, which may be a surface duct or a suspended duct. If M(z) suddenly increases at a height close to the ground, it may be a simple surface duct. According to the specific characteristics of the profile, if there is no atmospheric duct, it is marked as 0. If there is only one layer and it is a surface duct without a base layer, it is marked as 1. If there is only one layer and it is a surface duct with a base layer, it is marked as 2. If there is only one layer and it is a suspended duct, it is marked as 3. If it is a composite duct, it is marked as 4. Finally, a dataset containing the corrected atmospheric refractive index profile, longitude and latitude, time, duct bottom height, duct top height, duct thickness, duct strength, duct layer number, and duct type is constructed.

[0026] Furthermore, in S3, the method for matching AIS and meteorological data is as follows;

[0027] Step 1: Since the data reception density of the AIS base station is large per unit time, in this paper, the vessel information within 15 minutes before and after the whole hour is used. The time interval of the meteorological data is 1 hour. In order to match with the meteorological data, the time format of the AIS data is unified, the minute part is removed and classified into the adjacent whole hour time;

[0028] Step 2: The meteorological data is grid data with a density of 30km×30km. In order to associate the meteorological data with the AIS data, the ship position is normalized to the adjacent grid points with a spatial resolution of 30km×30km. Then it is regularized to the central grid point, and the grid point coordinates are the coordinates of the ship, and the meteorological data above the grid point is used as the meteorological data above the ship;

[0029] Step 3: After matching the AIS data and meteorological data according to time and location, divide the sea area near the AIS base station into 360 regions according to the azimuth angle, number each region from 1 to 360, calculate the azimuth angle between the grid point and the AIS base station and mark the region number, and then use the base station as the center of the circle, starting from a circular area 80 Km away from the center of the circle, divide multiple concentric circular regions at intervals of 10 Km;

[0030] Step 4: To ensure that the data is not interfered, remove the AIS position points where there is no atmospheric duct. Since the radio wave propagation in the AIS frequency band is only affected by the surface duct and suspended duct with a basic layer, the data with a ship-base distance greater than 80 Km is used for calculation. Therefore, it can be considered that the atmospheric ducts at the AIS position points are all surface ducts and suspended ducts with a basic layer.

[0031] Furthermore, in S4, use the adversarial production network to generate an atmospheric duct profile that can be used as the overall duct condition of the region according to the atmospheric duct profile at the ship position points where over-the-horizon propagation occurs in the region; if there is only 1 ship position point in the region per unit time, use its corresponding corrected atmospheric refractive index profile as the atmospheric duct profile of the region; if there are 2 or more ship position points in the region per unit time, then use the adversarial production network algorithm to calculate the new profile. The main principle of the algorithm is as follows: First, use the profile dataset real dataset as a sample, and at the same time randomly generate a noise vector from a Gaussian distribution. Then, input the noise vector into the generator to generate fake sample data. Finally, input the real sample and an equal amount of fake sample into the discriminator and use the mean square error for discrimination; use the simulated profile as the corrected atmospheric refractive index profile of the region.

[0032] Furthermore, the method for verifying the corrected atmospheric refractive index profile in S5 is as follows:

[0033] In the first step, substitute the corrected atmospheric refractive index profile above each ship position into the parabolic equation;

[0034]

[0035] x is the horizontal distance, z is the height, k = 2π / λ is the propagation constant in vacuum, m(x, z) is the corrected atmospheric refractive index, and the field component U(x, z) can be used to calculate the radio wave path loss Lp and the propagation factor Fp. The formulas are as follows;

[0036] L p (x, z) = -20lg|U(x, z)| + 20lg(4π) + 10lg(x) - 30lg(λ) (4)

[0037] F p (x, z) = 20lg|U(x, z)| + 10lg(x) + 10(λ) (5)

[0038] The propagation loss of the actual radio wave in the tropospheric atmosphere is;

[0039] L = L p - 20lgF p (6)

[0040] Table 1 AIS system parameters

[0041]

[0042] Finally, the actual propagation loss x of the radio wave corresponding to each ship within each unit time is constructed [x 1i , x 2i , …, x mi . The heights of the receiving base station and the transmitter are both taken as 10 m, so the propagation losses are all the data at a height of 10 m;

[0043] In the second step, the corrected atmospheric refractive index profile obtained by fitting in S4 is used to replace the profile corresponding to each ship, and the simulated propagation loss y of the radio wave corresponding to the ship is calculated by the parabolic equation [y 1i , y 2i , …, y i . The heights of the receiving base station and the transmitter are both taken as 10 m, so the propagation losses are all the data at a height of 10 m;

[0044] In the third step, by the Euclidean distance method, the similarity of each propagation loss curve in x[x 1i , x 2i , …, x mi and y[y 1i , y 2i , …, y mi is calculated. The formula is as follows:

[0045]

[0046] m is the number of profiles in each area, and i is the length of the propagation loss array. If all Ed are less than 0.5, it can be considered that this profile can represent the corrected refractive atmospheric refractive index profile at this time and in this area. Otherwise, return to S4 for recalculation.

[0047] In the fourth step, the optimal solution is output.

[0048] The beneficial effects of the present invention are as follows: The present invention uses the VHF signals emitted by the AIS system affected by the suspended waveguide and the surface waveguide with a base layer to invert the characteristic quantities of the suspended waveguide and the surface waveguide with a base layer, overcoming the problem that the traditional waveguide inversion method is easily affected by the evaporation duct, and obtaining more "clean" waveguide data; at the same time, due to the large number of AIS systems actually installed and the high spatio-temporal resolution of the signals, it can provide a richer data source for the inversion system, providing data support for building an atmospheric duct monitoring network using domestic AIS base stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments.

[0050] Figure 1 is the flowchart of the operation of the present invention;

[0051] Figure 2 is the flowchart of the AIS data screening and classification of the present invention;

[0052] Figure 3 is the flowchart of the profile modeling and waveguide type classification of the present invention;

[0053] Figure 4 is the flowchart of the time and space normalization of the AIS data of the present invention;

[0054] Figure 5 is the flowchart of the atmospheric data assimilation of the present invention;

[0055] Figure 6 is the flowchart of the dataset area division of the present invention;

[0056] Figure 7 is the schematic diagram of the area division of the present invention;

[0057] Figure 8 is the flowchart of the atmospheric profile fitting of the present invention;

[0058] Figure 9 is the flowchart of the verification of the fitted profile of the present invention. SPECIFIC IMPLEMENTATION SCHEMES

[0059] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the drawings of the present invention.

[0060] Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0061] As Figure 1 shown, the present invention proposes a new method for extracting regional atmospheric ducts based on AIS signals, and the method is as follows:

[0062] S1: Obtain AIS record data, and perform screening and classification processing on the data;

[0063] S2: Obtain atmospheric environment data, and construct a corrected atmospheric refractive index data set with a certain spatial resolution;

[0064] S3: Construct a data set of atmospheric corrected refractive index profiles associated with AIS signals, generate an AIS signal propagation environment database using AIS position information and atmospheric corrected refractive index profiles, screen the data, and divide the data set into regions;

[0065] S4: Use a generative adversarial network to establish a generator and a discriminator. The generator is used to construct a profile that is closest to the corrected atmospheric refractive index profile in the existing AIS signal propagation environment data set in the region, and the discriminator is responsible for judging whether the fitting result is accurate;

[0066] S5: Verify the fitting profile result. According to the position information recorded in the AIS in the data set, use the new profile and the original profile respectively, and calculate the power loss of radio wave propagation using the parabolic equation. Compare the similarity of the two power loss curves by the Euclidean distance method. If the calculation result of the Euclidean distance method for the loss curves at each point recorded by each AIS in the region is less than 0.5, then regard this new profile as the atmospheric duct result of the corresponding region.

[0067] The following embodiments explain the method in detail. The specific new method for extracting regional atmospheric ducts based on AIS signals is as follows:

[0068] S1: Process AIS data. The processing and screening of AIS data are as shown in Figure 2 shown;

[0069] S11: Process the AIS cipher, and obtain the recorded Coordinated Universal Time and the longitude and latitude of the ship's position;

[0070] S12: Calculate the distance between the ship and the base station according to the longitude and latitude. Since the normal propagation distance of AIS signals is 80 km, if the recorded ship-base distance is greater than 80 km, it must be affected by the atmospheric duct to increase the signal propagation distance. The data with a distance greater than 80 km can be marked as the data affected by the atmospheric duct in the signal propagation, and the data with a distance less than 80 km is marked as the data not affected by the atmospheric duct in the signal propagation process;

[0071]

[0072] Calculate the distance L between the ship and the AIS base station according to the longitude and latitude information i As shown in formula (1), where, (W i , W j ) and (Ji , J j ) are the longitude and latitude of the propagation position recorded by AIS and the receiving base station, R 地 The place is the radius of the earth, and the average value can be taken as 6371 km.

[0073] The decoding algorithm can be completed by using the existing technology.

[0074] S2: Obtain the atmospheric environment data and construct the dataset of the modified refractive index profile of the atmospheric environment. The process is as Figure 3 shown;

[0075] S21: Calculate the modified atmospheric refractive index profile from the temperature, pressure, and water vapor pressure data of the lower atmosphere of the ERA5 meteorological data, and identify and record the characteristic parameters of the atmospheric duct through interpolation. The formula is as follows:

[0076]

[0077] In the formula, M is the atmospheric modified refractive index exponent, which is a dimensionless quantity and is usually denoted as the M unit; T is the absolute temperature of the atmosphere (K); P is the atmospheric pressure (hPa); e is the water vapor pressure (hPa); z is the altitude (m).

[0078] S22: Classify the atmospheric duct situation. Calculate according to the calculation result M. When , it is considered that an atmospheric duct phenomenon occurs at the height of M n-1 . When , the atmospheric duct phenomenon ends at the height of M n-1 . M n-1 and M n are two atmospheric refractive indices at adjacent heights in the calculation result of the vertical distribution of the modified atmospheric refractive index. M n-1 is the atmospheric refractive index at the higher point, M n is the atmospheric refractive index at the lower point, and Δz is the height difference between the above adjacent heights. Therefore, M(z) increases with the decrease of height within a certain height range, indicating that there is a duct layer at this height, which may be a surface duct or a suspended duct. If M(z) suddenly increases at a height close to the ground, it may be a simple surface duct. According to the specific characteristics of the profile, if there is no atmospheric duct, it is marked as 0; if there is only one layer and it is a surface duct without a base layer, it is marked as 1; if there is only one layer and it is a surface duct with a base layer, it is marked as 2; if there is only one layer and it is a suspended duct, it is marked as 3; if it is a composite duct, it is marked as 4. Finally, construct a dataset containing the modified atmospheric refractive index profile, longitude and latitude, time, duct bottom height, duct top height, duct thickness, duct strength, duct layer number, and duct type;

[0079] Among them, the interpolation algorithm can be completed by using the existing technology.

[0080] S3: Construct a dataset of atmospheric modified refractive index profiles associated with AIS signals. The process of the AIS processing part is as Figure 4 shown.

[0081] S31: Since the data reception density of the AIS base station per unit time is large, in this embodiment, the vessel information within 15 minutes before and after the whole hour is used. The time interval of the meteorological data is 1 hour. To match the meteorological data, the time format of the AIS data is unified, the minute part is removed and classified into the adjacent whole hour time;

[0082] S32: The meteorological data is grid data with a density of 30km×30km. To associate the meteorological data with the AIS data, the vessel positions are normalized to the adjacent grid points with a spatial resolution of 30km×30km. Draw a square area with a side length of 30km centered on the grid point. If there is an AIS position record in the area, it is regularized to the central grid point, and the grid point coordinates are the coordinates of the vessel, and the meteorological data at the grid point is used as the meteorological data above the vessel;

[0083] S33: As Figure 5 shown, match the AIS data and the meteorological data according to time and position. As Figure 6 shown, calculate the azimuth angle between the meteorological grid point and the AIS base station. Divide the sea area near the AIS base station into 360 regions, calculate the azimuth angle between the grid point and the AIS base station and mark the region numbers, and then start from a circular area with a radius of 80Km centered on the base station, and divide multiple concentric circular areas at intervals of 10Km;

[0084] S34: To ensure that the data is not interfered, it is necessary to remove the AIS position points where there is no atmospheric duct. Since the radio wave propagation in the AIS frequency band is only affected by the surface duct and the suspended duct with a basic layer, the data with a ship-base distance greater than 80Km is used for calculation, and it can be considered that the atmospheric ducts at the AIS position points are all surface ducts and suspended ducts with a basic layer.

[0085] S4: The fitting process of the atmospheric modified refractive index profile is as Figure 8 shown.

[0086] S41: If there is only 1 vessel position point in the area per unit time, use the corresponding modified atmospheric refractive index profile as the atmospheric duct profile of the area;

[0087] S42: If there are two or more ship positions in the area per unit time, the adversarial production network algorithm is used to calculate the new profile. The main principle of the algorithm is as follows: First, use the profile dataset and the real dataset as samples, and at the same time randomly generate noise vectors from the Gaussian distribution. Then, input the noise vectors into the generator to generate fake sample data. Finally, input the real samples and an equal amount of fake samples into the discriminator and use the mean square error for discrimination; use the simulated profile as the corrected atmospheric refractive index profile for this area.

[0088] S5: Conduct the evaluation of the fitting profile results. The process is as Figure 9 shown.

[0089] S51: Substitute the corrected atmospheric refractive index profile above each ship position into the parabolic equation;

[0090]

[0091] x is the horizontal distance, z is the height, k = 2π / λ is the propagation constant in vacuum, m(x, z) is the corrected atmospheric refractive index. The field component U(x, z) can be obtained to calculate the radio wave path loss Lp and the propagation factor Fp. The formulas are as follows:

[0092] L p (x, z) = -20lg|U(x, z)| + 20lg(4π) + 10lg(x) - 30lg(λ) (4)

[0093] F p (x, z) = 20lg|U(x, z)| + 10lg(x) + 10(λ) (5)

[0094] The actual propagation loss of radio waves in the tropospheric atmosphere is:

[0095] L = L p - 20lgF p (6)

[0096] Table 2 AIS system parameters

[0097]

[0098] Finally, construct the actual propagation loss x[x 1i , x 2i , …, x mi of each ship corresponding to each unit time. The heights of the receiving base station and the transmitter are both taken as 10m. Therefore, the propagation loss is the data at a height of 10m;

[0099] S52: Use the corrected atmospheric refractive index profile obtained by fitting in S4 to replace the profile corresponding to each ship, and calculate the simulated propagation loss y[y1i , y 2i , …, y i , the heights of both the receiving base station and the transmitter are taken as 10 m, so the propagation losses are all data at a height of 10 m;

[0100] S53: By using the Euclidean distance method, calculate the similarity of each propagation loss curve between x[x 1i , x 2i , …, x mi and y[y 1i , y 2i , …, y mi . The formula is as follows:

[0101]

[0102] m is the number of profiles in each area, and i is the length of the propagation loss array. If all Ed are less than 0.5, it can be considered that this profile can represent the modified refractive atmospheric index profile at this time and in this area. Otherwise, return to S4 for recalculation;

[0103] S54: Output the optimal solution.

[0104] The present invention uses the very high frequency signals emitted by the AIS system to invert the characteristics of the elevated duct, overcomes the problem that the traditional duct inversion algorithm is easily affected by evaporation and surface ducts, and obtains purer elevated duct data; at the same time, due to the large installation volume of the AIS system and the high spatio-temporal resolution of the signals, it can provide rich data sources for duct inversion, overcomes the problem of single duct inversion results in the traditional algorithm, and improves the spatio-temporal resolution of the inverted duct area.

[0105] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A new method for extracting regional atmospheric waveguide based on AIS signals, characterized in that: include: S1: Obtain AIS record data, filter and classify the data; S2: Acquire atmospheric environment data and construct a corrected atmospheric refractive index dataset with a certain spatial resolution; S3: Construct a dataset of atmospheric corrected refractive index profiles associated with AIS signals, generate an AIS signal propagation environment dataset using AIS location information and atmospheric corrected refractive index profiles, filter the data, and classify the dataset into regions; The steps to construct a dataset of atmospheric corrected refractivity profiles associated with AIS signals and match AIS with meteorological data are as follows; S31 standardizes the time format of AIS data, removes the minute part and classifies it into the nearest hour time; S32 meteorological data is grid data with a density of 30km×30km. The grid points are drawn as square areas with a side length of 30km in the center. If there are AIS position records in the area, they will be regularized to the central grid points. The grid point coordinates are the coordinates of the ship, and the meteorological data above the grid points are used as the meteorological data above the ship. S33 matches the AIS data with the meteorological data according to time and location, divides the sea area near the AIS base station into 360 areas according to the azimuth, numbers each area from 1 to 360, calculates the azimuth between the grid point and the AIS base station and marks the area number, and divides the area into multiple concentric circle areas with a distance of 10 km, starting from a circular area 80 km from the center of the circle with the base station as the center; S34 uses data with a ship-to-base distance greater than 80 km for calculation, assuming that the atmospheric waveguides at the AIS position points at this distance are all surface waveguides and suspended waveguides with a foundation layer; S4: Establish a generative adversarial network, using a generator and a discriminator. The generator is used to construct a profile that is most similar to the modified atmospheric refractivity profile in the existing AIS signal propagation environment dataset in the area, and the discriminator is used to determine whether the fitting result is accurate, so as to obtain a more accurate atmospheric environment simulation result. In S4, the adversarial production network is used to generate the atmospheric waveguide profile as the overall waveguide condition of the area according to the atmospheric waveguide profile at the ship position point where over-the-horizon propagation occurs in the area. The specific steps are as follows; S41 If there is only one ship point in the area per unit time, the corresponding corrected atmospheric refractivity profile is used as the atmospheric waveguide profile of the area; S42 If there are two or more ship points in the area per unit time, the adversarial production network algorithm is used to calculate the new profile. The main principle of the algorithm is: first, the real data set of the profile data set is used as a sample, and a noise vector is randomly generated from a Gaussian distribution. Then, the noise vector is input into the generator to generate false sample data. Finally, the real sample and the same amount of false samples are input into the discriminator and the mean square error is used for discrimination; the simulated profile is used as the corrected atmospheric refractive index profile of the area; S5: Verify the simulation results of the fitted profile in S4. According to the location information of the AIS records in the data set, the new profile and the original profile are used respectively, and the parabolic equation is used to calculate the power loss of radio wave propagation; the similarity of the two power loss curves is compared by the Euclidean distance method; if the Euclidean distance calculation results of the loss curve at each AIS recorded point in the area are less than 0.5, the new profile is regarded as the atmospheric waveguide result of the corresponding area.

2. The novel method for extracting regional atmospheric waveguide based on AIS signal according to claim 1 is characterized in that: In step S1, target data is obtained, which specifically includes: Step 1: Process the AIS code to obtain the recorded universal time, longitude and latitude of the ship's position; Step 2: Calculate the distance between the ship and the signal receiving base station according to the longitude and latitude. Data with a distance greater than 80 km can be marked as data where the signal propagation is affected by the atmospheric duct, and data with a distance less than 80 km can be marked as data where the signal propagation process is not affected by the atmospheric duct. The latitude and longitude information is used to calculate the distance L between the ship and the AIS base station i As shown in formula (1), where (W i ,W j ) and (J i ,J j ) is the propagation position recorded by AIS and the longitude and latitude of the receiving base station, R 地 Earth is the radius of the earth, and the average value can be taken as 6371km.

3. The novel method for extracting regional atmospheric waveguide based on AIS signal according to claim 1 is characterized in that: In S2, the atmospheric environment is constructed by using the hourly ERA5 reanalysis meteorological data of the European Centre for Medium-Range Weather Forecasts reanalysis data to model the profile. The specific method is: Step 1: Use the hourly ERA5 meteorological reanalysis data released by the European Centre for Short-term Weather Forecasts. The horizontal resolution of the meteorological data is 0.25°. Use the temperature, air pressure, and water vapor pressure data of the lowest 12 atmospheric pressure layers of the ERA5 reanalysis data to calculate the corrected atmospheric refractivity profile, and identify and record the characteristic parameters of the suspended waveguide through interpolation. Step 2 maps the refractive index profile to grid points; Step 3: classify the surface waveguide, suspended waveguide and no waveguide according to the profile characteristics; Step 4 finally constructs a data set containing the corrected atmospheric refractive index profile, longitude and latitude, time, waveguide bottom height, waveguide top height, waveguide thickness, waveguide strength, waveguide layer number, and waveguide type.

4. The novel method for extracting regional atmospheric waveguide based on AIS signals according to claim 1, characterized in that: In S5; Step 1: Substitute the corrected atmospheric refractivity profile above each ship's position into the parabolic equation; x is the horizontal distance, z is the height, k = 2π / λ is the propagation constant in vacuum, m(x, z) is the corrected atmospheric refractive index, and the field component U(x, z) can be used to calculate the radio wave path loss Lp and propagation factor Fp. The formula is as follows: L p (x,z)=-20lg|U(x,z)|+20lg(4π)+10lg(x)-30lg(λ) (4) F p (x,z)=20lg|U(x,z)|+10lg(x)+10(λ) (5) The actual propagation loss of radio waves in the troposphere is: L=L p -20 lgF p (6) Finally, the actual propagation loss x[x 1i , x 2i , …, x mi ]; Step 2: Use the corrected atmospheric refractivity profile obtained by fitting in S4 to replace the profile corresponding to each ship, and calculate the radio wave simulation propagation loss y[y 1i ,y 2i , …, y i ]; Step 3: Calculate x[x 1i , x 2i , …, x mi ] and y[y 1i ,y 2i , …, y mi ]; the formula is as follows: m is the number of profiles in each region, and i is the length of the propagation loss array; if Ed is less than 0.5, it can be considered that the profile can represent the modified refraction atmospheric refractive index profile in the time region, otherwise return to S4 to recalculate; Step 4 outputs the optimal solution.

Images