Global real-time troposphere vertical correction method and device based on deep learning

Abstract

The invention provides a global real-time troposphere vertical correction method and device based on deep learning, and the method comprises the steps: obtaining multi-year troposphere delay data according to global atmosphere reanalysis data; establishing a troposphere vertical correction model by adopting a convolutional neural network, wherein the input of the troposphere vertical correction model comprises an annual day, the longitude and latitude of the current position, the height difference between the target position and the current position and the troposphere zenith delay; extracting spatio-temporal characteristic parameters through a convolutional layer, a pooling layer and a full connection layer, and verifying the precision of the troposphere vertical correction model by using MERRA-2 atmospheric reanalysis data as a reference value; the corresponding grid window is matched according to the current position information of the user, the corrected troposphere zenith delay value is output, and the model can be used for correcting the troposphere delay effect in navigation and positioning of a global navigation satellite system. The method is of great significance in providing real-time high-precision troposphere delay information of any position for global satellite navigation and positioning.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of satellite navigation and positioning technology, and in particular to a global real-time tropospheric vertical correction method and device based on deep learning. Background Art

[0002] The tropospheric delay effect has always been a key factor restricting the precision navigation and positioning and application efficiency of space technologies such as the Global Navigation Satellite System (GNSS) and Very Long Baseline Interferometry (VLBI). At present, most of the commonly used tropospheric vertical correction models in the lower atmosphere are mathematical models, which are difficult to accurately characterize the complex nonlinear changes in the tropospheric vertical profile. In addition, there are two extreme cases in the parameter setting of the existing global tropospheric vertical correction model: one is too many parameters (stored in the form of high-resolution grid points), and the other is too few parameters (globally unified parameters). Within a certain spatial range, too many model parameters not only cannot effectively improve the performance of tropospheric vertical correction, but may increase the computational burden; while too few parameters make it difficult to finely characterize the spatiotemporal changes of the global tropospheric vertical profile, which ultimately affects the vertical correction effect. With the rapid development of high-precision global services of global satellite navigation systems, it is urgent to develop new high-precision global tropospheric delay correction models and methods to meet the fine correction of tropospheric delay effects in precision navigation and positioning and applications of space technologies such as GNSS. Therefore, developing a global real-time tropospheric vertical correction method and equipment based on deep learning, which can effectively overcome the defects in the above-mentioned related technologies, has become a technical problem that needs to be urgently solved in the industry. Summary of the invention

[0003] In view of the above-mentioned problems existing in the prior art, an embodiment of the present invention provides a global real-time tropospheric vertical correction method and device based on deep learning.

[0004] In the first aspect, an embodiment of the present invention provides a global real-time tropospheric vertical correction method based on deep learning, comprising: obtaining multi-year tropospheric delay data according to the global atmospheric reanalysis data ERA5, wherein the time resolution of the tropospheric delay data is 1 hour and the spatial resolution is 0.25°×0.25°; establishing a tropospheric vertical correction model using a convolutional neural network, wherein the input of the tropospheric vertical correction model includes the annual accumulation day, the longitude and latitude of the current position, the height difference between the target position and the current position, and the tropospheric zenith delay ZTD; extracting spatiotemporal feature parameters through convolutional layers, pooling layers, and fully connected layers, and using MERRA-2 atmospheric reanalysis data as a reference value to verify the accuracy of the convolutional neural network tropospheric vertical correction model; matching the corresponding grid window according to the user's current position information, calling the convolutional neural network tropospheric vertical correction model, and outputting the corrected tropospheric zenith delay ZTD value, which can be used to correct the tropospheric delay effect in the navigation and positioning of the global navigation satellite system.

[0005] Based on the content of the above method embodiments, in the embodiments of the present invention, for the global real-time tropospheric vertical correction method based on deep learning, the acquisition of the tropospheric zenith delay ZTD includes:

[0006]

[0007] N = k1×(P - e) / T + k2×e / T + k3×e / T 2

[0008] e = h×P / 0.622

[0009] wherein, the tropospheric zenith delay amount ZTD is composed of ZTD1 and ZTD2. ZTD1 is the main tropospheric zenith delay; ZTD2 is the residual tropospheric zenith delay; N is the total atmospheric refractivity; n is the number of layers included in the atmospheric reanalysis data; the first refractivity coefficient k1 = 77.604 Kelvin per hectopascal; the second refractivity coefficient k2 = 64.79 Kelvin per hectopascal; the third refractivity coefficient k3 = 377600 Kelvin squared per hectopascal; H is the elevation; dH is the elevation change amount; P is the air pressure; T is the temperature; e is the water vapor pressure; h is the specific humidity; φ is the latitude corresponding to the global navigation satellite system site; N i and ΔH i are respectively the total atmospheric refractivity and the atmospheric thickness of the i-th layer; P top , e top , T top and H top are respectively the air pressure, water vapor pressure, temperature and elevation corresponding to the top layer of the atmospheric reanalysis data ERA5.

[0010] Based on the content of the above method embodiments, in the embodiments of the present invention, for the global real-time tropospheric vertical correction method based on deep learning, the tropospheric zenith delay amount ZTD is composed of ZTD1 and ZTD2, including: ZTD = ZTD1 + ZTD2.

[0011] Based on the content of the above method embodiments, in the embodiments of the present invention, for the global real-time tropospheric vertical correction method based on deep learning, obtaining the multi-year tropospheric delay data includes: using the integration method to divide the multi-year tropospheric delay data into grid points. The window size of the grid point division is 1°×1°, and the step size is 0.5°, to ensure the window continuity and the solvability of the model parameters.

[0012] Based on the content of the above method embodiments, the global real-time tropospheric vertical correction method based on deep learning provided in the embodiments of the present invention, the input layer of the convolutional neural network model includes 5 neurons, corresponding to the day of the year, the longitude and latitude of the current location, the height difference between the target location and the current location, and the tropospheric zenith delay ZTD; the convolutional layer uses a 3×3 convolutional kernel to extract spatial features, the pooling layer reduces the dimension through 2×2 max pooling, and the fully connected layer outputs the corrected tropospheric zenith delay ZTD value.

[0013] Based on the content of the above method embodiments, the global real-time tropospheric vertical correction method based on deep learning provided in the embodiments of the present invention, the convolutional neural network tropospheric vertical correction model can be used to correct the tropospheric delay effect in global navigation satellite system navigation positioning, and further includes: by comparing the results of the GGZTD-CNN model with the global classic GPT3 model, it is verified that the correction deviation of the GGZTD-CNN model is less than 2.5 cm.

[0014] Based on the content of the above method embodiments, the global real-time tropospheric vertical correction method based on deep learning provided in the embodiments of the present invention, after ensuring the window continuity and the solvability of the model parameters, further includes: the matching of the user's current location and the grid window is realized by dividing the longitude and latitude range, and the model parameters within the window include the spatio-temporal feature mapping relationship of longitude, latitude, day of the year, and height difference.

[0015] In a second aspect, an embodiment of the present invention provides a global real-time tropospheric vertical correction device based on deep learning, including: a first main module for obtaining multi-year tropospheric delay data according to the global atmospheric reanalysis data ERA5, the time resolution of the tropospheric delay data is 1 hour, and the spatial resolution is 0.25°×0.25°; a second main module for establishing a tropospheric vertical correction model using a convolutional neural network, the input of the tropospheric vertical correction model includes the day of the year, the longitude and latitude of the current location, the height difference between the target location and the current location, and the tropospheric zenith delay ZTD; a third main module for extracting spatio-temporal feature parameters through a convolutional layer, a pooling layer, and a fully connected layer, and using the MERRA-2 atmospheric reanalysis data as a reference value to verify the accuracy of the convolutional neural network tropospheric vertical correction model; a fourth main module for matching the corresponding grid window according to the user's current location information, calling the convolutional neural network tropospheric vertical correction model, and outputting the corrected tropospheric zenith delay ZTD value, and this model can be used to correct the tropospheric delay effect in global navigation satellite system navigation positioning.

[0016] In a third aspect, an embodiment of the present invention provides an electronic device, including:

[0017] At least one processor, at least one memory, and a communication interface; wherein,

[0018] The processor, the memory, and the communication interface communicate with each other;

[0019] The memory stores program instructions executable by the processor, and the processor calls the program instructions to execute the global real-time tropospheric vertical correction method based on deep learning provided by any one of the various implementation manners of the first aspect.

[0020] In a fourth aspect, an embodiment of the present invention provides a non-transitory computer-readable storage medium. The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause a computer to execute the global real-time tropospheric vertical correction method based on deep learning provided by any one of the various implementation manners of the first aspect.

[0021] The global real-time tropospheric vertical correction method and device provided by the embodiments of the present invention use the high-precision and high-resolution ECMWF fifth-generation global atmospheric reanalysis data ERA5 as the data source, and adopt a deep learning algorithm to establish a real-time high-precision global tropospheric vertical correction deep learning model, which solves the problems of fine characterization of the complex non-linear changes in the vertical profile of the lower atmosphere troposphere and the optimization of model parameters, and is of great significance for global satellite navigation and positioning and providing real-time high-precision tropospheric delay information at any location for atmospheric sounding. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for description in the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0023] Figure 1 It is a schematic flowchart of the global real-time tropospheric vertical correction method based on deep learning provided by the embodiment of the present invention;

[0024] Figure 2 It is a schematic structural diagram of the global real-time tropospheric vertical correction device based on deep learning provided by the embodiment of the present invention;

[0025] Figure 3 It is a schematic physical structure diagram of the electronic device provided by the embodiment of the present invention;

[0026] Figure 4 It is a schematic diagram of the accuracy verification and comparison effect of the tropospheric vertical correction model and the GPT3 model constructed by using a convolutional neural network (CNN) provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the protection scope of the present invention. In addition, the technical features in each embodiment or a single embodiment provided by the present invention can be combined with each other arbitrarily to form a feasible technical solution. Such combination is not restricted by the order of steps and / or the structural composition mode, but must be based on what can be achieved by those of ordinary skill in the art. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the protection scope required by the present invention. If there are step numbers in the following embodiments, they are only set for the convenience of elaboration and explanation, and no limitation is imposed on the order between steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0028] An embodiment of the present invention provides a global real-time tropospheric vertical correction method based on deep learning. Refer to Figure 1 , the method includes: obtaining multi-year tropospheric delay data according to the global atmospheric reanalysis data ERA5, where the time resolution of the tropospheric delay data is 1 hour and the spatial resolution is 0.25°×0.25°; establishing a tropospheric vertical correction model using a convolutional neural network, where the input of the tropospheric vertical correction model includes the day of the year, the longitude and latitude of the current position, the height difference between the target position and the current position, and the tropospheric zenith delay ZTD; extracting spatio-temporal feature parameters through convolutional layers, pooling layers, and fully connected layers, and using the MERRA-2 atmospheric reanalysis data as a reference value to verify the accuracy of the convolutional neural network tropospheric vertical correction model; matching the corresponding grid window according to the user's current position information, calling the convolutional neural network tropospheric vertical correction model, and outputting the corrected tropospheric zenith delay ZTD value, which can be used to correct the tropospheric delay effect in global navigation satellite system navigation and positioning.

[0029] A global real-time tropospheric vertical correction method based on deep learning provided by an embodiment of the present invention is directed to GNSS atmospheric correction services. Using the tropospheric delay data provided by the ERA5 atmospheric reanalysis data, a tropospheric vertical correction model suitable for the lower atmosphere is established using the convolutional neural network (CNN) algorithm. Users can find the corresponding window according to the position information, and use the time information and combine the model parameters of the corresponding window to achieve the vertical correction of the tropospheric delay effect at the user's position, which is of great significance for improving the performance of GNSS tropospheric delay atmospheric services and high-precision applications.

[0030] Based on the content of the above method embodiments, as an alternative embodiment, in the embodiments of the present invention, for the global real-time tropospheric vertical correction method based on deep learning, the acquisition of the tropospheric zenith delay ZTD includes:

[0031]

[0032] N = k1×(P - e) / T + k2×e / T + k3×e / T 2

[0033] e = h×P / 0.622

[0034] Among them, the tropospheric zenith delay amount ZTD consists of ZTD1 and ZTD2. ZTD1 is the main tropospheric zenith delay; ZTD2 is the residual tropospheric zenith delay; N is the total atmospheric refractivity; n is the number of layers included in the atmospheric reanalysis data; the first refractivity coefficient k1 = 77.604 Kelvin per hectopascal; the second refractivity coefficient k2 = 64.79 Kelvin per hectopascal; the third refractivity coefficient k3 = 377600 Kelvin squared per hectopascal; H is the elevation; dH is the elevation change amount; P is the atmospheric pressure; T is the temperature; e is the water vapor pressure; h is the specific humidity; φ is the latitude corresponding to the global navigation satellite system site; N i and ΔH i are respectively the total atmospheric refractivity and the atmospheric thickness of the i-th layer; P top , e top , T top and H top are respectively the atmospheric pressure, water vapor pressure, temperature and elevation corresponding to the top layer of the atmospheric reanalysis data ERA5.

[0035] Based on the content of the above method embodiments, as an alternative embodiment, in the embodiments of the present invention, for the global real-time tropospheric vertical correction method based on deep learning, the tropospheric zenith delay amount ZTD consists of ZTD1 and ZTD2, and includes: ZTD = ZTD1 + ZTD2.

[0036] Specifically, a tropospheric vertical correction model is constructed based on the multi-year tropospheric delay data provided by the high-precision and high-resolution ECMWF fifth-generation global atmospheric reanalysis data (ERA5); the advantage of the deep learning algorithm (CNN) in characterizing the complex non-linear variation relationship between variables is proposed, and a real-time high-precision global tropospheric vertical correction deep learning model is established to solve the problems of fine characterization of the complex non-linear variation of the tropospheric vertical profile in the lower atmosphere and the optimization of model parameters. Among them, the integral method is used to calculate the tropospheric ZTD values of different isobaric layers at each grid point, and the calculation formula is as shown in ZTD.

[0037] Based on the content of the above method embodiments, as an alternative embodiment, in the global real-time tropospheric vertical correction method based on deep learning provided in the embodiments of the present invention, the obtaining of the multi-year tropospheric delay data includes: dividing the multi-year tropospheric delay data into grid points by using the integral method, where the window size of the grid point division is 1°×1°, and the step size is 0.5°, to ensure the window continuity and the solvability of the model parameters. The key to grid dissection is to determine the window size and step size thereof. The determination of the grid size needs to consider principles such as the integer nature of the dissection number, the continuity of the window, and the solvability of the model parameters within the window.

[0038] Based on the content of the above method embodiments, as an alternative embodiment, in the global real-time tropospheric vertical correction method based on deep learning provided in the embodiments of the present invention, the input layer of the convolutional neural network model includes 5 neurons, corresponding to the day of the year, the longitude and latitude of the current position, the height difference between the target position and the current position, and the tropospheric zenith delay ZTD respectively; the convolutional layer uses a 3×3 convolutional kernel to extract spatial features, the pooling layer reduces the dimension through 2×2 max pooling, and the fully connected layer outputs the corrected tropospheric zenith delay ZTD value.

[0039] Specifically, the convolutional neural network (CNN) has strong complex non-linear representation performance and has unique advantages in capturing the spatio-temporal characteristic parameters of tropospheric delay. This deep learning method for global real-time precise tropospheric vertical correction can solve the optimization problem of model parameters. Specifically, reference can be made to Figure 4 , taking the MERRA-2 data as the reference value, verifying the accuracy of the tropospheric vertical correction model and the GPT3 model constructed by using the convolutional neural network (CNN), so as to show that the deep learning algorithm is capable of capturing the complex non-linear changes in the vertical profile of the lower atmosphere troposphere.

[0040] Based on the content of the above method embodiments, as an alternative embodiment, in the global real-time tropospheric vertical correction method based on deep learning provided in the embodiments of the present invention, the convolutional neural network tropospheric vertical correction model can be used to correct the tropospheric delay effect in global navigation satellite system navigation positioning, and further includes: by comparing the results of the GGZTD-CNN model with the global classic GPT3 model, verifying that the correction deviation of the GGZTD-CNN model is less than 2.5 cm.

[0041] Based on the content of the above method embodiments, as an alternative embodiment, in the global real-time tropospheric vertical correction method based on deep learning provided in the embodiments of the present invention, after ensuring the window continuity and the solvability of the model parameters, it further includes: the matching between the user's current position and the grid window is realized through the division of longitude and latitude ranges, and the model parameters within the window include the spatio-temporal characteristic mapping relationships of longitude, latitude, day of the year, and height difference.

[0042] Specifically, the data preprocessing part needs to input the day of year (Doy), the ZTD of the user's location, longitude (Lon), latitude (Lat), and the height difference between the target location and the current location; the convolutional neural network part mainly includes an input layer, a convolutional layer, a pooling layer, and a fully connected layer, which are used to extract features through convolution, reduce the resolution through pooling, and summarize features respectively; the output part is mainly used to output the ZTD after vertical correction by the convolutional neural network algorithm. The convolutional neural network (CNN) has strong complex non-linear characterization performance and has unique advantages in capturing the spatio-temporal characteristic parameters of tropospheric delay. This deep learning method for global real-time precise tropospheric vertical correction can solve the optimization problem of model parameters.

[0043] The global real-time tropospheric vertical correction method based on deep learning provided by the embodiments of the present invention uses the high-precision and high-resolution ECMWF fifth-generation global atmospheric reanalysis data ERA5 as the data source, and establishes a real-time high-precision global tropospheric vertical correction deep learning model by using deep learning algorithms, to solve the problems of fine characterization of complex non-linear changes in the vertical profile of the lower atmosphere troposphere and the optimization of model parameters, which is of great significance for global satellite navigation and positioning and providing real-time high-precision tropospheric delay information at any location for atmospheric sounding.

[0044] The implementation basis of each embodiment of the present invention is achieved through programmed processing by a device with processor functions. Therefore, in engineering practice, the technical solutions and their functions of each embodiment of the present invention can be encapsulated into various modules. Based on this actual situation, on the basis of the above embodiments, the embodiments of the present invention provide a global real-time tropospheric vertical correction device based on deep learning, and this device is used to execute the global real-time tropospheric vertical correction method in the above method embodiments. See Figure 2 , this device includes: a first main module, which is used to obtain multi-year tropospheric delay data according to the global atmospheric reanalysis data ERA5, and the time resolution of the tropospheric delay data is 1 hour, and the spatial resolution is 0.25°×0.25°; a second main module, which is used to establish a tropospheric vertical correction model by using a convolutional neural network, and the input of the tropospheric vertical correction model includes the day of year, the longitude and latitude of the current location, the height difference between the target location and the current location, and the tropospheric zenith delay ZTD; a third main module, which is used to extract spatio-temporal characteristic parameters through a convolutional layer, a pooling layer, and a fully connected layer, and use the MERRA-2 atmospheric reanalysis data as a reference value to verify the accuracy of the convolutional neural network tropospheric vertical correction model; a fourth main module, which is used to match the corresponding grid window according to the user's current location information, call the convolutional neural network tropospheric vertical correction model, and output the corrected tropospheric zenith delay ZTD value, and this model can be used to correct the tropospheric delay effect in global navigation satellite system navigation and positioning.

[0045] The global real-time tropospheric vertical correction device based on deep learning provided by the embodiments of the present invention adopts Figure 2 several modules therein. By using the high-precision and high-resolution ECMWF fifth-generation global atmospheric reanalysis data ERA5 as the data source and establishing a real-time high-precision global tropospheric vertical correction deep learning model using deep learning algorithms, it solves the problems of accurately characterizing the complex non-linear variations in the vertical profile of the lower atmosphere troposphere and optimizing model parameters, which is of great significance for global satellite navigation and positioning and providing real-time high-precision tropospheric delay information at any location for atmospheric sounding.

[0046] It should be noted that the device in the device embodiment provided by the present invention can be used not only to implement the method in the above method embodiment, but also to implement the methods in other method embodiments provided by the present invention. The difference is only in setting corresponding functional modules. The principle is basically the same as that of the above device embodiment provided by the present invention. As long as those skilled in the art, on the basis of the above device embodiment, refer to the specific technical solutions in other method embodiments, obtain corresponding technical means by combining technical features, and the technical solutions composed of these technical means, and on the premise of ensuring the practicability of the technical solutions, the device in the above device embodiment can be improved to obtain corresponding device-type embodiments for implementing the methods in other method-type embodiments. For example:

[0047] Based on the content of the above device embodiment, as an optional embodiment, the global real-time tropospheric vertical correction device based on deep learning provided by the embodiments of the present invention further includes: a first sub-module for obtaining the tropospheric zenith delay ZTD, including:

[0048]

[0049] N = k1×(P - e) / T + k2×e / T + k3×e / T 2

[0050] e = h×P / 0.622

[0051] wherein, the tropospheric zenith delay ZTD is composed of ZTD1 and ZTD2. ZTD1 is the main tropospheric zenith delay; ZTD2 is the residual tropospheric zenith delay; N is the total atmospheric refractivity; n is the number of layers included in the atmospheric reanalysis data; the first refractivity coefficient k1 = 77.604 Kelvin per hectopascal; the second refractivity coefficient k2 = 64.79 Kelvin per hectopascal; the third refractivity coefficient k3 = 377600 Kelvin squared per hectopascal; H is the elevation; dH is the elevation change; P is the air pressure; T is the temperature; e is the water vapor pressure; h is the specific humidity; φ is the latitude corresponding to the global navigation satellite system site; Ni and ΔH i are respectively the total atmospheric refractive index and the atmospheric thickness of the i-th layer; P top , e top , T top and H top are respectively the air pressure, water vapor pressure, temperature and elevation corresponding to the top layer of the atmospheric reanalysis data ERA5.

[0052] Based on the content of the above device embodiments, as an alternative embodiment, the global real-time tropospheric vertical correction device based on deep learning provided in the embodiments of the present invention further includes: a second sub-module, configured to implement that the tropospheric zenith delay ZTD is composed of ZTD1 and ZTD2, including: ZTD = ZTD1 + ZTD2.

[0053] Based on the content of the above device embodiments, as an alternative embodiment, the global real-time tropospheric vertical correction device based on deep learning provided in the embodiments of the present invention further includes: a third sub-module, configured to implement obtaining multi-year tropospheric delay data, including: dividing the multi-year tropospheric delay data into grid points by using the integral method, the window size of the grid point division is 1°×1°, and the step size is 0.5°, to ensure the window continuity and the solvability of the model parameters.

[0054] Based on the content of the above device embodiments, as an alternative embodiment, the global real-time tropospheric vertical correction device based on deep learning provided in the embodiments of the present invention further includes: a fourth sub-module, configured to implement that the input layer of the convolutional neural network tropospheric vertical correction model includes 5 neurons, corresponding to the day of the year, the longitude and latitude of the current location, the height difference between the target location and the current location, and the tropospheric zenith delay ZTD; the convolutional layer uses a 3×3 convolutional kernel to extract spatial features, the pooling layer reduces the dimension through 2×2 max pooling, and the fully connected layer outputs the corrected tropospheric zenith delay ZTD value.

[0055] Based on the content of the above device embodiments, as an alternative embodiment, the global real-time tropospheric vertical correction device based on deep learning provided in the embodiments of the present invention further includes: a fifth sub-module, the convolutional neural network tropospheric vertical correction model can be used to correct the tropospheric delay effect in the global navigation satellite system navigation and positioning, and further includes: by comparing the results of the GGZTD-CNN model with the global classic GPT3 model, verifying that the correction deviation of the GGZTD-CNN model is less than 2.5 cm.

[0056] Based on the content of the above device embodiments, as an alternative embodiment, the device for global real-time tropospheric vertical correction based on deep learning provided in the embodiments of the present invention further includes: a sixth sub-module, which is used to implement the following after ensuring the continuity of the window and the solvability of the model parameters: the matching between the user's current location and the grid window is achieved through the division of longitude and latitude ranges, and the model parameters within the window include the spatio-temporal feature mapping relationships of longitude, latitude, day of the year, and height difference.

[0057] The method of the embodiments of the present invention is implemented relying on an electronic device. Therefore, it is necessary to introduce the relevant electronic device. For this purpose, the embodiments of the present invention provide an electronic device, as Figure 3 shown, the electronic device includes: at least one processor, a communication interface, at least one memory, and a communication bus. Among them, the at least one processor, the communication interface, and the at least one memory complete mutual communication through the communication bus. The at least one processor can call the logical instructions in the at least one memory to execute all or part of the steps of the methods provided in the foregoing method embodiments.

[0058] In addition, when the logical instructions in the above-mentioned at least one memory are implemented in the form of a software functional unit and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or this part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various method embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc that can store program codes.

[0059] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative labor.

[0060] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0061] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to multiple embodiments of the present invention. Based on this understanding, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and sometimes may be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0062] It should be noted that the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or device that includes a series of elements includes not only those elements but also other elements that are not explicitly listed, or elements that are inherent to such process, method, article, or device. Without further limitation, the elements defined by the statement "including..." do not exclude the existence of additional identical elements in the process, method, article, or device that includes the said elements. For any "predetermined threshold", "preset threshold", or similar expression, if no specific value is indicated, those of ordinary skill in the art can determine its specific value through simple experiments or corresponding debugging.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A global real-time tropospheric vertical correction method based on deep learning, characterized in that Including: Based on the global atmospheric reanalysis data ERA5, multi-year tropospheric delay data is obtained. The time resolution of the tropospheric delay data is 1 hour, and the spatial resolution is 0.25°×0.25°. A convolutional neural network is used to establish a tropospheric vertical correction model. The input of the tropospheric vertical correction model includes the day of the year, the longitude and latitude of the current position, the height difference between the target position and the current position, and the tropospheric zenith delay ZTD. Spatiotemporal feature parameters are extracted through convolutional layers, pooling layers, and fully connected layers. Using MERRA-2 atmospheric reanalysis data as a reference value, the accuracy of the convolutional neural network tropospheric vertical correction model is verified. According to the user's current position information, the corresponding grid window is matched, and the convolutional neural network tropospheric vertical correction model is called to output the corrected tropospheric zenith delay ZTD value. This model can be used to correct the tropospheric delay effect in global navigation satellite system navigation and positioning.

2. The global real-time tropospheric vertical correction method based on deep learning according to claim 1, characterized in that, The acquisition of the tropospheric zenith delay ZTD includes: N = k1×(P - e) / T + k2×e / T + k3×e / T 2 e = h×P / 0.622 Among them, the tropospheric zenith delay ZTD is composed of ZTD1 and ZTD2. ZTD1 is the main tropospheric zenith delay; ZTD2 is the residual tropospheric zenith delay; N is the total atmospheric refractivity; n is the number of layers included in the atmospheric reanalysis data; the first refractivity coefficient k1 = 77.604 Kelvin per hectopascal; the second refractivity coefficient k2 = 64.79 Kelvin per hectopascal; the third refractivity coefficient k3 = 377600 Kelvin squared per hectopascal; H is the elevation; dH is the elevation change; P is the air pressure; T is the temperature; e is the water vapor pressure; h is the specific humidity; φ is the latitude corresponding to the global navigation satellite system site; N i and ΔH i are the total atmospheric refractivity and the atmospheric thickness of the i-th layer respectively; P top , e top , T top and H top are the air pressure, water vapor pressure, temperature and elevation corresponding to the top layer of the atmospheric reanalysis data ERA5 respectively.

3. The global real-time tropospheric vertical correction method based on deep learning according to claim 2, wherein The tropospheric zenith delay amount ZTD is composed of ZTD1 and ZTD2, including: ZTD = ZTD1 + ZTD2.

4. The method for global real-time tropospheric vertical correction based on deep learning according to claim 3, characterized in that The obtaining of the multi-year tropospheric delay data includes: using the integration method to divide the multi-year tropospheric delay data into grid points. The window size of the grid point division is 1°×1°, and the step size is 0.5°, ensuring the window continuity and the solvability of the model parameters.

5. The global real-time tropospheric vertical correction method based on deep learning according to claim 4, characterized in that The input layer of the convolutional neural network model contains 5 neurons, corresponding to the day of the year, the longitude and latitude of the current position, the height difference between the target position and the current position, and the tropospheric zenith delay ZTD respectively. The convolutional layer uses a 3×3 convolutional kernel to extract spatial features, the pooling layer reduces the dimension through 2×2 max pooling, and the fully connected layer outputs the corrected tropospheric zenith delay ZTD value.

6. The global real-time tropospheric vertical correction method based on deep learning according to claim 5, wherein The convolutional neural network tropospheric vertical correction model can be used to correct the tropospheric delay effect in global navigation satellite system navigation and positioning. It also includes: by comparing the results of the GGZTD-CNN model with the global classical GPT3 model, it is verified that the correction deviation of the GGZTD-CNN model is less than 2.5 cm.

7. The global real-time tropospheric vertical correction method based on deep learning according to claim 6, wherein After ensuring the window continuity and the solvability of the model parameters, it also includes: the matching between the user's current position and the grid window is achieved through the division of longitude and latitude ranges. The model parameters within the window include the spatiotemporal feature mapping relationships of longitude, latitude, day of the year, and height difference.

8. A global real-time tropospheric vertical correction device based on deep learning, characterized in that, Including: The first main module is used to obtain multi-year tropospheric delay data based on the global atmospheric reanalysis data ERA5. The time resolution of the tropospheric delay data is 1 hour, and the spatial resolution is 0.25°×0.25°. The second main module is used to establish a tropospheric vertical correction model using a convolutional neural network. The input of the tropospheric vertical correction model includes the day of the year, the longitude and latitude of the current position, the height difference between the target position and the user's current position, and the tropospheric zenith delay ZTD. The third main module is used to extract spatio-temporal feature parameters through convolutional layers, pooling layers, and fully connected layers, and use the MERRA-2 atmospheric reanalysis data as a reference value to verify the accuracy of the convolutional neural network tropospheric vertical correction model. The fourth main module is used to match the corresponding grid window according to the user's current position information, call the convolutional neural network tropospheric vertical correction model, and output the corrected tropospheric zenith delay ZTD value. This model can be used to correct the tropospheric delay effect in global navigation satellite system navigation and positioning.

9. An electronic device, characterized in that, Comprising: At least one processor, at least one memory, and a communication interface; wherein, The processor, the memory, and the communication interface communicate with each other; The memory stores program instructions executable by the processor, and the processor calls the program instructions to execute the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the method according to any one of claims 1 to 7.

Images