A training and calculation method, device and equipment for ionospheric network model

Abstract

The present application provides a training and calculation method, device and equipment for an ionospheric network model, and relates to the field of ionospheric technology. The method obtains ionospheric calculation grid data, ionospheric calculation grid data and ionospheric label grid data of a preset unit time period of a target area; according to the ionospheric calculation grid data of the preset unit time period, an initial first network model is used for prediction to obtain the ionospheric initial prediction grid data of the target area; according to the ionospheric calculation grid data and the ionospheric initial prediction grid data, an initial second network model is used for prediction to obtain the ionospheric target prediction grid data of the target area; according to the ionospheric label grid data and the ionospheric target prediction grid data of the target area, the first network model and the second network model are trained to obtain the target first network model and the target second network model. Thus, a target model with high accuracy, high resolution and low latency is obtained through training.

Description

Technical Field

[0001] The present invention relates to the field of ionosphere technology, and in particular to a training and calculation method, device and equipment for an ionosphere network model. Background Art

[0002] As a widely distributed medium, the ionosphere may induce various ionospheric effects on radio waves passing through it, including reflection, scattering, absorption and refraction. These effects not only cause important errors in the positioning, navigation and timing services of the Global Navigation Satellite System (GNSS), but may also lead to space weather disasters such as radio communication interruption, charging of spacecraft surface and power system collapse, which have extremely adverse effects on human production and life.

[0003] Therefore, on the one hand, there is an urgent need to develop a high-precision empirical ionospheric model to meet the requirements of space weather research and be able to be applied to the correction of GNSS ionospheric delay; on the other hand, there is an urgent need to realize short-term warning and forecast of the ionosphere to provide timely warning services for human production and life.

[0004] In recent years, Convolutional Neural Network (CNN) has been introduced into the prediction modeling of ionospheric TEC due to its outstanding performance in the field of image data processing. However, in practical applications, the use of neural networks in modeling the ionosphere cannot meet the requirements of high precision, high resolution and real-time performance. Summary of the invention

[0005] The purpose of the present invention is to address the deficiencies in the above-mentioned prior art. The present application provides a training and calculation method, device and equipment for an ionospheric network model to solve the problems in the prior art that the high-precision, high-resolution and real-time performance cannot be achieved.

[0006] To achieve the above purpose, the technical solution adopted in the embodiment of the present application is as follows:

[0007] In a first aspect, an embodiment of the present application provides a method for training an ionospheric network model, wherein the ionospheric network model includes: a target first network model and a target second network model, and the method includes:

[0008] Acquire ionospheric grid training data of the target area, wherein the ionospheric grid training data includes: ionospheric calculation grid data of each area in a preset unit time period before each historical time point, ionospheric calculation grid data of each area at each historical time point, and ionospheric label grid data of each area at each historical time point;

[0009] According to the ionosphere calculation grid data of the preset unit time period before each historical time point, the initial first network model is used for prediction to obtain the initial ionosphere prediction grid data of the target area at each historical time point;

[0010] According to the ionospheric calculation grid data at each historical time point and the ionospheric initial prediction grid data at each historical time point, an initial second network model is used to perform prediction to obtain the ionospheric target prediction grid data of the target area at each historical time point;

[0011] The first network model and the second network model are trained according to the ionospheric label grid data of the target area at each historical time point and the ionospheric target prediction grid data to obtain the target first network model and the target second network model.

[0012] Optionally, the acquiring ionospheric grid training data of the target area includes:

[0013] Obtaining ionospheric signal observation values ​​of multiple stations in the target area at each historical time point;

[0014] According to the ionospheric signal observation values ​​of the multiple stations at the historical time points, an ionospheric grid data set of the target area is obtained, wherein the ionospheric grid data set of the target area includes: ionospheric calculation grid data of the target area at the historical time points and ionospheric label grid data of the target area at the historical time points;

[0015] According to the preset unit time, the ionospheric grid data set of the target area is segmented to obtain multiple groups of the ionospheric grid training data of the target area.

[0016] Optionally, obtaining the ionospheric grid dataset of the target area according to the ionospheric signal observation values ​​of the multiple stations at the historical time points includes:

[0017] Calculate the ionospheric TEC calculated values ​​of the multiple stations at the historical time points using a first data model according to the ionospheric signal observation values ​​of the multiple stations at the historical time points;

[0018] According to the ionospheric signal observation values ​​of the multiple stations at the historical time points, a second data model is used to calculate the ionospheric TEC label values ​​of the multiple stations at the historical time points, wherein the calculation accuracy of the second data model is higher than the calculation accuracy of the first data model.

[0019] Optionally, obtaining the ionospheric grid dataset of the target area according to the ionospheric signal observation values ​​of the multiple stations at the historical time points further includes:

[0020] The consistency of the ionospheric signal observation values ​​of the multiple stations at the historical time points and the ionospheric TEC label values ​​is verified.

[0021] Optionally, before calculating the ionospheric TEC calculated values ​​of the multiple stations at the historical time points using the first data model according to the ionospheric signal observation values ​​of the multiple stations at the historical time points, the method further includes:

[0022] According to the ionospheric signal observation values ​​of the multiple stations at the historical time points, mathematical modeling is performed using a non-difference and non-combination algorithm to obtain the first data model.

[0023] Optionally, segmenting the ionospheric grid data set of the target area according to the preset unit time to obtain multiple groups of the ionospheric grid training data of the target area includes:

[0024] The ionospheric grid data set of the target area is segmented by moving a moving time window of a preset duration to obtain multiple groups of ionospheric grid training data of the target area.

[0025] In a second aspect, an embodiment of the present application provides an ionospheric data calculation method, which is applied to an ionospheric network model trained by the ionospheric network model training method described in any one of the first aspects above, wherein the ionospheric network model includes: a target first network model and a target second network model, and the method includes:

[0026] Acquire the ionospheric grid data of the target area in a preset unit time period before the time to be calculated, and the ionospheric calculation grid data of the time to be calculated;

[0027] Using the target first network model, predicting the ionospheric grid data of the preset unit time period, and obtaining the ionospheric prediction grid data of the target area at the time to be calculated;

[0028] The target second network model is used to predict the ionospheric calculation grid data and the ionospheric prediction grid data at the time to be calculated, so as to obtain the ionospheric target grid data of the target area at the time to be calculated.

[0029] In a third aspect, an embodiment of the present application provides a training device for an ionospheric network model, wherein the ionospheric network model includes: a target first network model and a target second network model, and the device includes:

[0030] An acquisition module, used for acquiring ionospheric grid training data of a target area, including: ionospheric calculation grid data of each area in a preset unit time period before each historical time point, ionospheric calculation grid data of each area at each historical time point, and ionospheric label grid data of each area at each historical time point;

[0031] A first calculation module is used to calculate grid data of the ionosphere in the preset unit time period before each historical time point, and use an initial first network model to perform prediction to obtain initial predicted grid data of the ionosphere of the target area at each historical time point;

[0032] A second calculation module is used to calculate the ionosphere grid data at each historical time point and the ionosphere initial prediction grid data at each historical time point, and use the initial second network model to perform prediction to obtain the ionosphere target prediction grid data of the target area at each historical time point;

[0033] A training module is used to train the first network model and the second network model according to the ionospheric label grid data of the target area at each historical time point and the ionospheric target prediction grid data to obtain the target first network model and the target second network model.

[0034] In a fourth aspect, an embodiment of the present application provides an ionospheric data calculation device, which is applied to an ionospheric network model trained by the ionospheric network model training method described in any one of the first aspects above, wherein the ionospheric network model includes: a target first network model and a target second network model, and the device includes:

[0035] A second acquisition module is used to acquire the ionospheric grid data of the target area in a preset unit time period before the time to be calculated, and the ionospheric calculation grid data of the time to be calculated;

[0036] A third calculation module is used to use the target first network model to predict the ionospheric grid data of the preset unit time period to obtain the ionospheric prediction grid data of the target area at the time to be calculated;

[0037] The fourth calculation module is used to adopt the target second network model to predict the ionospheric calculation grid data and the ionospheric prediction grid data at the time to be calculated, so as to obtain the ionospheric target grid data of the target area at the time to be calculated.

[0038] In a fifth aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a storage medium, wherein the processor and the storage medium are communicatively connected via a bus, the storage medium stores program instructions executable by the processor, and the processor calls the program stored in the storage medium to execute the steps of the ionospheric network model training method as described in any one of the first aspects, or executes the steps of the ionospheric data calculation method as described in the second aspect.

[0039] Compared with the prior art, this application has the following beneficial effects:

[0040] The present application provides a training and calculation method, device and equipment for an ionospheric network model. The method obtains ionospheric grid training data of a target area, wherein the ionospheric grid training data includes: ionospheric calculation grid data of each area in a preset unit time period before each historical time point, ionospheric calculation grid data of each area at each historical time point, and ionospheric label grid data of each area at each historical time point; according to the ionospheric calculation grid data of the preset unit time period before each historical time point, an initial first network model is used for prediction to obtain the ionospheric initial prediction grid data of the target area at each historical time point; according to the ionospheric calculation grid data at each historical time point and the ionospheric initial prediction grid data at each historical time point, an initial second network model is used for prediction to obtain the ionospheric target prediction grid data of the target area at each historical time point; according to the ionospheric label grid data and the ionospheric target prediction grid data of the target area at each historical time point, the first network model and the second network model are trained to obtain the target first network model and the target second network model. Therefore, by referring to the ionosphere calculation grid data and the ionosphere label grid data, a target model with high accuracy, high resolution and low latency is trained. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0042] Figure 1 A flowchart of a training method for an ionospheric network model provided in this application;

[0043] Figure 2 A flowchart of a method for obtaining ionospheric grid training data of a target area provided in an embodiment of the present application;

[0044] Figure 3A flowchart of a method for obtaining an ionospheric grid dataset of a target area provided in an embodiment of the present application;

[0045] Figure 4 A schematic diagram of a flow chart of an ionospheric data calculation method provided in this application;

[0046] Figure 5 A schematic diagram of a training device for an ionospheric network model provided in an embodiment of the present application;

[0047] Figure 6 A schematic diagram of an ionospheric data calculation device provided in an embodiment of the present application;

[0048] Figure 7 A schematic diagram of an electronic device provided in an embodiment of the present application.

[0049] Icons: 501 - acquisition module, 502 - first calculation module, 503 - second calculation module, 504 - training module, 601 - second acquisition module, 602 - third calculation module, 603 - fourth calculation module, 701 - processor, 702 - storage medium. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0051] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for which protection is sought, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0052] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0053] In addition, the terms “first”, “second”, etc., if used, are merely used to distinguish between the descriptions and should not be understood as indicating or implying relative importance.

[0054] It should be noted that, in the absence of conflict, the features in the embodiments of the present invention may be combined with each other.

[0055] The following is an explanation of a training method for an ionospheric network model provided in this application through a specific example. Figure 1 The present invention provides a flow chart of a training method for an ionospheric network model. The execution subject of the method may be an electronic device, which may be a device with computing and processing functions, such as a desktop computer, a laptop computer, etc. Figure 1 As shown, the ionospheric network model includes: a target first network model and a target second network model, and the method includes:

[0056] S101. Obtain ionospheric grid training data of a target area.

[0057] The ionospheric grid training data includes: the ionospheric calculation grid data of each region in a preset unit time period before each historical time point, the ionospheric calculation grid data of each region at each historical time point, and the ionospheric label grid data of each region at each historical time point.

[0058] The ionosphere calculation grid data is the model input data, and the ionosphere label grid data is the label data for training the model. Both the ionosphere calculation grid data and the ionosphere label grid data are existing ionosphere grid data, and the accuracy of the ionosphere label grid data is greater than that of the ionosphere calculation grid data.

[0059] The ionospheric grid data is the TEC (Total Electron Content) values ​​of multiple stations in the ionosphere. The TEC values ​​of all stations together constitute the ionospheric grid data.

[0060] S102, calculating grid data of the ionosphere in a preset unit time period before each historical time point, using an initial first network model for prediction, and obtaining initial predicted grid data of the ionosphere in the target area at each historical time point.

[0061] For example, the first network model is a ConvLSTM model. The preset unit time period may be 24 hours. That is, the ionospheric calculation grid data 24 hours before the historical time point is input into the initial ConvLSTM model to obtain the ionospheric initial prediction grid data at the historical time point.

[0062] The ionospheric calculation grid data before the historical time point is used to predict the ionospheric grid data at the historical time point, so that the calculated ionospheric grid data at the historical time point refers to the previous ionospheric calculation grid data, so that the trained model can calculate the ionospheric grid data more accurately.

[0063] S103, according to the ionosphere calculation grid data at each historical time point and the ionosphere initial prediction grid data at each historical time point, the initial second network model is used for prediction to obtain the ionosphere target prediction grid data of the target area at each historical time point.

[0064] For example, the second network model is a CycleGAN model, which can meet the requirements of high precision, high resolution and real-time performance. The ionospheric calculation grid data at each historical time point obtained in S101 and the ionospheric initial prediction grid data at each historical time point predicted in S102 are input into the CycleGAN model as a data pair to obtain the ionospheric target prediction grid data of the target area at each historical time point.

[0065] The calculated target prediction grid data refers to the ionosphere calculation grid data and the ionosphere initial prediction grid data, so that the trained model can calculate the ionosphere grid data more accurately.

[0066] S104. According to the ionospheric label grid data of the target area at each historical time point and the ionospheric target prediction grid data, the first network model and the second network model are trained to obtain a target first network model and a target second network model.

[0067] After each training, the ionosphere label grid data and the ionosphere target prediction grid data are compared. If the ionosphere label grid data and the ionosphere target prediction grid data meet the preset training conditions, the current first network model and the second network model are used as the target first network model and the target second network model. If the ionosphere label grid data and the ionosphere target prediction grid data do not meet the preset training conditions, the model parameters of the first network model and the second network model are adjusted until the ionosphere label grid data and the ionosphere target prediction grid data meet the preset training conditions.

[0068] For example, the preset training condition is that the absolute value of the difference between the ionospheric label grid data and the ionospheric target prediction grid data, and the ratio of the absolute value of the difference between the ionospheric label grid data and the ionospheric target prediction grid data is less than a preset value, for example, the preset value is 5%.

[0069] Specifically, all data is divided into training set, validation set and test set in a ratio of 7:2:1 to complete model training. Thus, the target model with high accuracy and low latency is obtained through training.

[0070] In summary, in this embodiment, the ionospheric grid training data of the target area is obtained, wherein the ionospheric grid training data includes: the ionospheric calculation grid data of each area in a preset unit time period before each historical time point, the ionospheric calculation grid data of each area at each historical time point, and the ionospheric label grid data of each area at each historical time point; according to the ionospheric calculation grid data of the preset unit time period before each historical time point, the initial first network model is used for prediction to obtain the ionospheric initial prediction grid data of the target area at each historical time point; according to the ionospheric calculation grid data at each historical time point, and the ionospheric initial prediction grid data at each historical time point, the initial second network model is used for prediction to obtain the ionospheric target prediction grid data of the target area at each historical time point; according to the ionospheric label grid data and the ionospheric target prediction grid data of the target area at each historical time point, the first network model and the second network model are trained to obtain the target first network model and the target second network model. Therefore, by referring to the ionosphere calculation grid data and the ionosphere label grid data, a target model with high accuracy, high resolution and low latency is trained.

[0071] In the above Figure 1 On the basis of the corresponding embodiment, the embodiment of the present application also provides a method for obtaining ionospheric grid training data of a target area. Figure 2 A flowchart of a method for obtaining ionospheric grid training data of a target area provided in an embodiment of the present application. Figure 2 As shown, obtaining ionospheric grid training data of the target area in S101 includes:

[0072] S201. Obtain ionospheric signal observation values ​​of multiple stations in the target area at various historical time points. Collect long-term and continuous GNSS (Global Navigation Satellite System) observation data from receiving stations located in the target area (or the target area and surrounding areas within a preset range), and ensure that the data sources are extensive and comprehensive. Organize the collected GNSS observation data into the RINEX format to ensure data consistency and integrity.

[0073] By checking the signal-to-noise ratio of the sensor of the acquisition equipment, the equipment status and other indicators, the observations with poor quality are excluded to improve the reliability of the data. The observations with large gross errors are eliminated, and by comparing the differences between the observations and the surrounding values, the extreme values ​​that may be abnormal are excluded to ensure the accuracy of the data; statistical methods or correlation analysis are used to exclude observations with strong correlations to prevent data redundancy from affecting the analysis results; and the pre-processed observation data are integrated for subsequent comprehensive analysis.

[0074] S202. Obtain an ionospheric grid data set of a target area according to ionospheric signal observation values ​​of multiple stations at various historical time points.

[0075] The ionospheric grid data set of the target area includes: the ionospheric calculation grid data of the target area at each historical time point and the ionospheric label grid data of the target area at each historical time point.

[0076] S203. Segment the ionospheric grid data set of the target area according to a preset unit time to obtain multiple groups of ionospheric grid training data of the target area.

[0077] In order to facilitate subsequent model training, the ionospheric grid data set of the target area is segmented to obtain multiple groups of ionospheric grid training data. In each round of training, a group of ionospheric grid training data is input.

[0078] In summary, in this embodiment, the ionospheric signal observation values ​​of multiple stations in the target area at each historical time point are obtained; according to the ionospheric signal observation values ​​of multiple stations at each historical time point, the ionospheric grid data set of the target area is obtained, and the ionospheric grid data set of the target area includes: the ionospheric calculation grid data of the target area at each historical time point and the ionospheric label grid data of the target area at each historical time point; according to the preset unit time, the ionospheric grid data set of the target area is segmented to obtain multiple groups of ionospheric grid training data of the target area. Thus, multiple groups of ionospheric grid training data of the target area are accurately obtained.

[0079] In the above Figure 2 On the basis of the corresponding embodiments, the embodiments of the present application also provide a method for obtaining an ionospheric grid data set of a target area. Figure 3 A flow chart of a method for obtaining an ionospheric grid dataset of a target area provided in an embodiment of the present application. Figure 3 As shown, in S202, according to the ionospheric signal observation values ​​of multiple stations at each historical time point, an ionospheric grid data set of the target area is obtained, including:

[0080] S301. Calculate ionospheric TEC calculated values ​​of the plurality of stations at each historical time point using a first data model according to ionospheric signal observation values ​​of the plurality of stations at each historical time point.

[0081] The ionospheric signal observation value of each station at each historical time point is input into the first data model in turn, and the ionospheric TEC calculated value of each station at each historical time point is calculated.

[0082] S302. According to the ionospheric signal observation values ​​of the multiple stations at each historical time point, the second data model is used to calculate the ionospheric TEC label values ​​of the multiple stations at each historical time point.

[0083] The ionospheric signal observation value of each station at each historical time point is input into the second data model in turn, and the ionospheric TEC label value of each station at each historical time point is calculated.

[0084] The calculation accuracy of the second data model is higher than that of the first data model.

[0085] For example, the second data model is the International Reference Ionosphere (IRI) model.

[0086] The calculated ionospheric TEC values ​​and ionospheric TEC label values ​​at each historical time point are stored in a format suitable for subsequent model training and convenient for processing and analysis.

[0087] In summary, in this embodiment, according to the ionospheric signal observation values ​​of multiple stations at each historical time point, the first data model is used to calculate the ionospheric TEC calculation values ​​of multiple stations at each historical time point; according to the ionospheric signal observation values ​​of multiple stations at each historical time point, the second data model is used to calculate the ionospheric TEC label values ​​of multiple stations at each historical time point, wherein the calculation accuracy of the second data model is higher than the calculation accuracy of the first data model. Thus, each ionospheric TEC calculation value and ionospheric TEC label value are accurately calculated.

[0088] In the above Figure 3 On the basis of the corresponding embodiment, in another embodiment of the present application, in S202, the ionospheric grid data set of the target area is obtained according to the ionospheric signal observation values ​​of multiple stations at each historical time point, and further includes:

[0089] The consistency of the ionospheric signal observation values ​​and ionospheric TEC label values ​​of multiple stations at various historical time points is verified.

[0090] First, ensure that the actual ionospheric signal observation value and the ionospheric TEC tag value are in the same time range. Match the spatial position of the ionospheric signal observation value and the ionospheric TEC tag value to ensure that the geographical coordinates of the comparison points are consistent.

[0091] Then, the error between the ionospheric TEC tag value and the ionospheric signal observation value is calculated. The indicator used is the root mean square error. The specific calculation method is shown in the following formula (1):

[0092]

[0093] where y i is the ionospheric TEC tag value, is the ionospheric signal observation value.

[0094] Summarize the error index results and distribution comparison, conduct comparative analysis, comprehensively evaluate the accuracy and reliability of the model, store qualified data, and ensure that the results are high-quality and reliable. Such labeled data sets are used as label results for model training to improve the accuracy and generalization ability of the built model for the ionosphere.

[0095] In summary, in this embodiment, the consistency of the ionospheric signal observation values ​​of multiple stations at various historical time points and the ionospheric TEC label values ​​is verified, thereby building the accuracy and generalization ability of the ionosphere model.

[0096] In the above Figure 3 On the basis of the corresponding embodiment, in another embodiment of the present application, before calculating the ionospheric TEC calculated values ​​of the multiple stations at each historical time point using the first data model according to the ionospheric signal observation values ​​of the multiple stations at each historical time point in S301, the method further includes:

[0097] According to the ionospheric signal observation values ​​of multiple stations at various historical time points, mathematical modeling is performed using a non-difference and non-combination algorithm to obtain a first data model.

[0098] Specifically, according to the basic observation equations of pseudorange and phase of GNSS data, the ionospheric parameters can be regarded as unknown parameters. For the sake of simplicity, the tropospheric error, relativistic effect, multipath error, etc. are combined, which can be expressed as the following formula (2):

[0099]

[0100] in, It represents the distance from the receiver r to the satellite s, including orbit error, tropospheric error, relativistic effect, multipath error, etc. They represent floating point ambiguities including satellite and receiving phase hardware delay bias respectively; They represent the pseudorange and phase observations of satellite s at the ith frequency observed by receiver r; c represents the propagation speed of light in vacuum; dt r 、dt s represents the clock difference of receiver r and satellite s; B r,i , represents the pseudorange hardware delay between the receiver and the satellite at the ith frequency; represents the observation noise and unmodeled error. represents the ionospheric delay on the observation path from receiver r to satellite s.

[0101] When performing PPP (non-differential non-combined) solution, the precise satellite clock error products released by IGS are used for correction. However, the values ​​released by IGS are based on the ionospheric-free pseudorange and phase observation values, resulting in the satellite clock error containing the satellite hardware delay bias, which can be expressed as the following formula (3):

[0102]

[0103] in, Indicates the true value of the satellite clock error; dt s is the IGS published value of the satellite clock error; are the frequency values ​​of two signals with different frequencies; These are the satellite hardware delays for the two frequencies respectively.

[0104] In the non-combined PPP model, the satellite hardware delay bias cannot be offset in the pseudorange observation value, but can be absorbed by the ionospheric parameters; in the phase observation value, it can be absorbed by the ionospheric parameters and the ambiguity parameters. Therefore, by substituting formula (3) into formula (2) and reorganizing the parameters, the following formula (4) can be obtained:

[0105]

[0106] The specific expressions of each symbol in formula (4) are shown in formula (5):

[0107]

[0108] in, They represent the pseudorange and phase observation values ​​of satellite s observed by receiver r at the i-th frequency respectively; represents the distance from the receiver r to the satellite s; and represents the receiver and satellite clock difference after the equation parameters are reorganized; B r,i , represents the pseudorange hardware delay between the receiver and the satellite at the ith frequency; represents the ionospheric delay on the observation path from receiver r to satellite s, including the receiver and satellite hardware bias; represents the ionospheric delay on the observation path from receiver r to satellite s excluding receiver and satellite hardware bias; are the frequency values ​​of two signals with different frequencies; They represent floating point ambiguities including satellite and receiving phase hardware delay bias respectively; They represent the floating point ambiguities without satellite and receiving phase hardware delay bias respectively; Represents observation noise and unmodeled errors; DCBr,i and Represents the differential code deviation between different frequencies; γ is the proportional coefficient.

[0109] Among them, there is formula (6) as follows:

[0110]

[0111] Will By estimating the parameters, the ionospheric delay deviation on the slant path can be solved, and then the actual ionospheric observation value can be obtained.

[0112] Then, based on the thin ionosphere assumption, the following projection function (MSLM, Modified SingleLayer Model) is used to calculate the ionosphere mapping function, thereby converting the oblique path ionosphere delay to the vertical ionosphere delay. The calculation formula is shown in the following formula (7):

[0113]

[0114] Where z' is the zenith distance in the direction of the signal path at the puncture point; z is the zenith distance at the receiver; R is the distance between the receiver and the center of the earth; H is the height of the thin layer of the ionosphere, which is generally a constant at a certain height of 350-450km.

[0115] The regional ionosphere is modeled by combining the spherical harmonics model, the ionospheric delay obtained by the UPPP algorithm, and the ionospheric mapping function. Specifically, the ionospheric delay obtained by the UPPP algorithm is converted into the ionospheric delay in the vertical direction using the ionospheric mapping function, and then input into the spherical harmonics model to calculate the ionospheric TEC value.

[0116] In summary, in this embodiment, according to the ionospheric signal observation values ​​of multiple stations at each historical time point, a non-difference non-combination algorithm is used to perform mathematical modeling to obtain a first data model. Thus, the first data model is accurately obtained.

[0117] In the above Figure 2 On the basis of the corresponding embodiment, in S203, the ionospheric grid data set of the target area is segmented according to the preset unit time to obtain multiple groups of ionospheric grid training data of the target area, including:

[0118] The ionospheric grid dataset of the target area is segmented using a moving time window of preset duration to obtain multiple groups of ionospheric grid training data of the target area.

[0119] For example, the preset time length is 24 hours, and the moving segmentation method is to segment each movement by a sliding window of 24 hours. That is, each set of ionospheric grid training data includes: ionospheric calculation grid data 24 hours before the historical time point, and ionospheric calculation grid data at the historical time point.

[0120] In summary, in this embodiment, a moving time window of a preset duration is used to perform moving segmentation on the ionospheric grid data set of the target area to obtain multiple sets of ionospheric grid training data of the target area. Thus, multiple sets of ionospheric grid training data of the target area are accurately obtained.

[0121] The following is an explanation of a method for calculating ionospheric data provided in the present application through a specific example. Figure 4 The present invention provides a flowchart of a method for calculating ionospheric data. The method may be performed by an electronic device, which may be a device with computing and processing functions, such as a desktop computer, a laptop computer, etc. Figure 4 As shown, the ionospheric network model obtained by training using the training method of the ionospheric network model of any of the above embodiments, the ionospheric network model includes: a target first network model and a target second network model, and the method includes:

[0122] S401, obtaining ionospheric grid data of a target area in a preset unit time period before a time to be calculated, and ionospheric calculation grid data of the time to be calculated.

[0123] For example, the preset unit time period is 24 hours. The ionospheric calculation grid data is pre-calculated and stored for subsequent use. For specific calculation methods, refer to Figure 2 The corresponding embodiment diagram, S301 in the embodiment 3 and the corresponding embodiments below it will not be described in detail here.

[0124] S402: using the target first network model, predicting the ionospheric grid data of a preset unit time period, and obtaining the ionospheric prediction grid data of the target area at the time to be calculated.

[0125] For example, the target first network model is a trained ConvLSTM model. The preset unit time period can be 24 hours. That is, the ionospheric calculation grid data 24 hours before the historical time point is input into the initial ConvLSTM model to obtain the ionospheric prediction grid data of the target area at the time to be calculated.

[0126] The ionosphere calculation grid data before the historical time point is used to predict the ionosphere grid data at the historical time point, so that the calculated ionosphere grid data at the historical time point refers to the previous ionosphere calculation grid data, and the ionosphere grid data is calculated more accurately.

[0127] S403, using the target second network model, predicting the ionosphere calculation grid data at the time to be calculated and the ionosphere prediction grid data, to obtain the ionosphere target grid data of the target area at the time to be calculated.

[0128] For example, the target second network model is a trained CycleGAN model, which can meet the requirements of high precision, high resolution and real-time performance. The ionosphere calculation grid data of the time to be calculated obtained in S401 and the ionosphere prediction grid data of the time to be calculated obtained in S402 are input into the CycleGAN model as a data pair to obtain the ionosphere target grid data of the time to be calculated.

[0129] The calculated ionospheric target grid data refers to the ionospheric calculation grid data and the ionospheric prediction grid data, so that the ionospheric target grid data can be calculated more accurately.

[0130] In summary, in this embodiment, the ionospheric grid data of the target area in a preset unit time period before the time to be calculated and the ionospheric calculation grid data at the time to be calculated are obtained; the target first network model is used to predict the ionospheric grid data of the preset unit time period to obtain the ionospheric prediction grid data of the target area at the time to be calculated; the target second network model is used to predict the ionospheric calculation grid data at the time to be calculated and the ionospheric prediction grid data to obtain the ionospheric target grid data of the target area at the time to be calculated. Thus, a model with high accuracy and low delay is used, and the ionospheric calculation grid data and the ionospheric prediction grid data are referred to to calculate more accurate ionospheric target grid data.

[0131] The following is a description of the training of an ionospheric network model, an ionospheric data calculation device, equipment, and storage medium provided by the present application. The specific implementation process and technical effects are described above and will not be repeated below.

[0132] Figure 5 A schematic diagram of a training device for an ionospheric network model provided in an embodiment of the present application, such as Figure 5 As shown, the ionospheric network model includes: a target first network model and a target second network model, and the device includes:

[0133] The acquisition module 501 is used to acquire the ionospheric grid training data of the target area, including: the ionospheric calculation grid data of each area in a preset unit time period before each historical time point, the ionospheric calculation grid data of each area at each historical time point, and the ionospheric label grid data of each area at each historical time point.

[0134] The first calculation module 502 is used to calculate the grid data of the ionosphere in a preset unit time period before each historical time point, and use the initial first network model to perform prediction to obtain the initial predicted grid data of the ionosphere in the target area at each historical time point.

[0135] The second calculation module 503 is used to calculate the ionosphere grid data at each historical time point and the initial prediction grid data of the ionosphere at each historical time point, and use the initial second network model to perform prediction to obtain the ionosphere target prediction grid data of the target area at each historical time point.

[0136] The training module 504 is used to train the first network model and the second network model according to the ionospheric label grid data and the ionospheric target prediction grid data of the target area at each historical time point to obtain the target first network model and the target second network model.

[0137] Furthermore, the acquisition module 501 is specifically used to obtain ionospheric signal observation values ​​of multiple stations in the target area at each historical time point; according to the ionospheric signal observation values ​​of the multiple stations at each historical time point, an ionospheric grid data set of the target area is obtained, and the ionospheric grid data set of the target area includes: ionospheric calculation grid data of the target area at each historical time point and ionospheric label grid data of the target area at each historical time point; according to a preset unit time, the ionospheric grid data set of the target area is segmented to obtain multiple groups of ionospheric grid training data of the target area.

[0138] Furthermore, the acquisition module 501 is specifically used to calculate the ionospheric TEC calculation values ​​of the multiple stations at each historical time point using a first data model based on the ionospheric signal observation values ​​of the multiple stations at each historical time point; and calculate the ionospheric TEC label values ​​of the multiple stations at each historical time point using a second data model based on the ionospheric signal observation values ​​of the multiple stations at each historical time point, wherein the calculation accuracy of the second data model is higher than the calculation accuracy of the first data model.

[0139] Furthermore, the acquisition module 501 is specifically used to verify the consistency of the ionospheric signal observation values ​​of multiple stations at each historical time point and the ionospheric TEC label value.

[0140] Furthermore, the acquisition module 501 is specifically used to perform mathematical modeling using a non-difference and non-combination algorithm according to the ionospheric signal observation values ​​of multiple stations at various historical time points to obtain a first data model.

[0141] Furthermore, the acquisition module 501 is specifically used to perform mobile segmentation on the ionospheric grid data sets of multiple stations using a moving time window of a preset duration to obtain multiple groups of ionospheric grid training data of the multiple stations.

[0142] Figure 6 A schematic diagram of an ionospheric data calculation device provided in an embodiment of the present application, such as Figure 6 As shown, the ionospheric network model obtained by training using the training method of the ionospheric network model of any of the above embodiments, the ionospheric network model includes: a target first network model and a target second network model, and the device includes:

[0143] The second acquisition module 601 is used to acquire the ionospheric grid data of the target area in a preset unit time period before the time to be calculated, and the ionospheric calculation grid data of the time to be calculated.

[0144] The third calculation module 602 is used to use the target first network model to predict the ionospheric grid data of a preset unit time period to obtain the ionospheric prediction grid data of the target area at the time to be calculated.

[0145] The fourth calculation module 603 is used to use the target second network model to predict the ionosphere calculation grid data at the time to be calculated and the ionosphere prediction grid data to obtain the ionosphere target grid data of the target area at the time to be calculated.

[0146] Figure 7 A schematic diagram of an electronic device provided in an embodiment of the present application, which electronic device may be a device with computing and processing functions.

[0147] The electronic device includes: a processor 701 and a storage medium 702. The processor 701 and the storage medium 702 are connected via a bus.

[0148] The storage medium 702 is used to store programs, and the processor 701 calls the programs stored in the storage medium 702 to execute the above method embodiment. The specific implementation method and technical effect are similar and will not be repeated here.

[0149] Optionally, the present invention also provides a storage medium, including a program, which is used to execute the above-mentioned method embodiment when executed by a processor. In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0150] 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 on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0151] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0152] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (English: processor) to perform some steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (English: Read-Only Memory, abbreviated: ROM), random access memory (English: Random Access Memory, abbreviated: RAM), disk or optical disk and other media that can store program codes.

Claims

1. A training method for an ionospheric network model, characterized in that: The ionospheric network model includes: a target first network model and a target second network model, and the method includes: Acquire ionospheric grid training data of the target area, wherein the ionospheric grid training data includes: ionospheric calculation grid data of each area in a preset unit time period before each historical time point, ionospheric calculation grid data of each area at each historical time point, and ionospheric label grid data of each area at each historical time point; According to the ionosphere calculation grid data of the preset unit time period before each historical time point, the initial first network model is used for prediction to obtain the initial ionosphere prediction grid data of the target area at each historical time point; According to the ionospheric calculation grid data at each historical time point and the ionospheric initial prediction grid data at each historical time point, an initial second network model is used to perform prediction to obtain the ionospheric target prediction grid data of the target area at each historical time point; The first network model and the second network model are trained according to the ionospheric label grid data of the target area at each historical time point and the ionospheric target prediction grid data to obtain the target first network model and the target second network model.

2. The method according to claim 1, characterized in that The step of acquiring ionospheric grid training data of the target area comprises: Obtaining ionospheric signal observation values ​​of multiple stations in the target area at each historical time point; According to the ionospheric signal observation values ​​of the multiple stations at the historical time points, an ionospheric grid data set of the target area is obtained, wherein the ionospheric grid data set of the target area includes: ionospheric calculation grid data of the target area at the historical time points and ionospheric label grid data of the target area at the historical time points; According to the preset unit time, the ionospheric grid data set of the target area is segmented to obtain multiple groups of the ionospheric grid training data of the target area.

3. The method according to claim 2, characterized in that The step of obtaining the ionospheric grid dataset of the target area according to the ionospheric signal observation values ​​of the multiple stations at the historical time points includes: Calculate the ionospheric TEC calculated values ​​of the multiple stations at the historical time points using a first data model according to the ionospheric signal observation values ​​of the multiple stations at the historical time points; According to the ionospheric signal observation values ​​of the multiple stations at the historical time points, a second data model is used to calculate the ionospheric TEC label values ​​of the multiple stations at the historical time points, wherein the calculation accuracy of the second data model is higher than the calculation accuracy of the first data model.

4. The method according to claim 3, characterized in that The step of obtaining the ionospheric grid data set of the target area according to the ionospheric signal observation values ​​of the multiple stations at the historical time points further includes: The consistency of the ionospheric signal observation values ​​of the multiple stations at the historical time points and the ionospheric TEC label values ​​is verified.

5. The method according to claim 3, characterized in that: Before calculating the ionospheric TEC calculated values ​​of the multiple stations at the historical time points using the first data model according to the ionospheric signal observation values ​​of the multiple stations at the historical time points, the method further includes: According to the ionospheric signal observation values ​​of the multiple stations at the historical time points, mathematical modeling is performed using a non-difference and non-combination algorithm to obtain the first data model.

6. The method according to claim 2, characterized in that The step of segmenting the ionospheric grid data set of the target area according to the preset unit time to obtain multiple groups of the ionospheric grid training data of the target area includes: The ionospheric grid data set of the target area is segmented by moving a moving time window of a preset duration to obtain multiple groups of ionospheric grid training data of the target area.

7. A method for calculating ionospheric data, characterized in that: The ionospheric network model is applied to an ionospheric network model trained by the training method of the ionospheric network model described in any one of claims 1 to 6, wherein the ionospheric network model comprises: a target first network model and a target second network model, and the method comprises: Acquire the ionospheric grid data of the target area in a preset unit time period before the time to be calculated, and the ionospheric calculation grid data of the time to be calculated; Using the target first network model, predicting the ionospheric grid data of the preset unit time period, and obtaining the ionospheric prediction grid data of the target area at the time to be calculated; The target second network model is used to predict the ionospheric calculation grid data and the ionospheric prediction grid data at the time to be calculated, so as to obtain the ionospheric target grid data of the target area at the time to be calculated.

8. A training device for an ionospheric network model, characterized in that: The ionospheric network model includes: a target first network model and a target second network model, and the device includes: An acquisition module, used for acquiring ionospheric grid training data of a target area, including: ionospheric calculation grid data of each area in a preset unit time period before each historical time point, ionospheric calculation grid data of each area at each historical time point, and ionospheric label grid data of each area at each historical time point; A first calculation module is used to calculate grid data of the ionosphere in the preset unit time period before each historical time point, and use an initial first network model to perform prediction to obtain initial predicted grid data of the ionosphere of the target area at each historical time point; A second calculation module is used to calculate the ionosphere grid data at each historical time point and the ionosphere initial prediction grid data at each historical time point, and use the initial second network model to perform prediction to obtain the ionosphere target prediction grid data of the target area at each historical time point; A training module is used to train the first network model and the second network model according to the ionospheric label grid data of the target area at each historical time point and the ionospheric target prediction grid data to obtain the target first network model and the target second network model.

9. An ionospheric data calculation device, characterized in that: The ionospheric network model is applied to an ionospheric network model trained by the training method of the ionospheric network model described in any one of claims 1 to 6, wherein the ionospheric network model comprises: a target first network model and a target second network model, and the device comprises: A second acquisition module is used to acquire the ionospheric grid data of the target area in a preset unit time period before the time to be calculated, and the ionospheric calculation grid data of the time to be calculated; A third calculation module is used to use the target first network model to predict the ionospheric grid data of the preset unit time period to obtain the ionospheric prediction grid data of the target area at the time to be calculated; The fourth calculation module is used to adopt the target second network model to predict the ionospheric calculation grid data and the ionospheric prediction grid data at the time to be calculated, so as to obtain the ionospheric target grid data of the target area at the time to be calculated.

10. An electronic device, characterized in that: include: A processor and a storage medium, wherein the processor and the storage medium are communicatively connected via a bus, the storage medium stores program instructions executable by the processor, and the processor calls the program stored in the storage medium to execute the steps of the ionospheric network model training method according to any one of claims 1 to 6, or executes the steps of the ionospheric data calculation method according to claim 7.

Images