A Multi-Device Systematic Error Correction Method for Ionospheric Total Electron Content

Abstract

The present invention discloses a multi-device systematic error correction method for ionospheric total electron content, including: acquiring dual-frequency observables collected by GNSS ground-based observation stations in the research area, solving the slant total electron content STEC, and obtaining GNSS_VTEC; reading the atmospheric density profiles collected by COSMIC satellites to obtain COSMIC_VTEC, integrating and processing the GNSS_VTEC and COSMIC_VTEC data within the selected time period and saving them; registering the GNSS_VTEC within a preset time range and a limited geographical location range through the processed COSMIC_VTEC, inputting it into a BP neural network for processing, and outputting the VTEC prediction difference; correcting the COSMIC_VTEC according to the VTEC prediction difference to obtain the corrected data. The present invention can effectively and efficiently correct the observables of space-based COSMIC.

Description

Technical Field

[0001] The present invention relates to the technical field of ionospheric correction in satellite navigation, and particularly to a multi-device system error correction method for ionospheric total electron content. Background Art

[0002] The ionosphere is an ionized region in the upper atmosphere of the Earth, filled with a large number of free electrons and ions, which has a significant impact on the propagation of radio signals. Since humans discovered the reflection of high-frequency radio waves by the ionosphere in the early 20th century, scientists have realized that the changes in the ionosphere will affect the transmission quality of signals. Due to the dynamic changes of the ionosphere under the action of space weather events (such as solar wind and geomagnetic storms), the precise monitoring and modeling of the ionosphere have become the core requirements for space weather monitoring and improving the accuracy of navigation and positioning and the stability of communication systems.

[0003] Ionospheric monitoring and modeling based on multi-source data contribute to improving the reconstruction accuracy and resolution of the ionosphere. However, different monitoring systems have different systematic errors, which limit the consistency and accuracy of the data. The COSMIC (Constellation Observing System for Meteorology, Ionosphere, and Climate) occultation project obtains global atmosphere and ionosphere data all day and all weather through radio occultation (RO) of the Global Navigation Satellite System (GNSS), with the advantage of wide coverage. It effectively makes up for the problem of insufficient coverage of detection networks such as ground-based GNSS, ionosondes, and incoherent scatter radars, and can especially obtain ionospheric vertical distribution data. Research shows that in the ionospheric vertical distribution data of the current COSMIC-2 occultation project, the unqualified rate of data near the equator is about 25%, while it is reduced to about 15% in the mid-low latitude regions. In addition, the comparison between the ionospheric vertical total electron content (VTEC) data obtained by the occultation technique and the ground-based ionospheric vertical total electron content (GNSS_VTEC) data shows that the root mean square error (RMSE) is between 4 and 6 TECU. In order to improve the accuracy of the ionospheric model based on multi-source data, it is necessary to correct the systematic errors between multi-source data. Existing correction methods, such as the least squares method and support vector machines, rely on weighted regression and model verification, are sensitive to data quality and assumption conditions, and it is difficult to comprehensively improve the accuracy of the data. To address these problems, the present invention proposes a multi-source data correction method based on a multi-layer backpropagation neural network (BP) to achieve more accurate error compensation and data correction. Summary of the Invention

[0004] Aiming at the poor quality of ionospheric occultation data and the deficiencies of traditional correction methods, the present invention proposes a multi-device systematic error correction method for the total electron content in the ionosphere, providing an accurate data source for global high-precision and high spatio-temporal resolution ionospheric modeling.

[0005] To achieve the above object, the present invention provides a multi-device systematic error correction method for the total electron content in the ionosphere, including:

[0006] Obtain the dual-frequency observables collected by GNSS ground-based observation stations in the study area, use the carrier-smoothed pseudorange method to solve the slant total electron content STEC on the line of sight from the ground-based observation station to the satellite, and use the mapping function to calculate and obtain the vertical total electron content of the ground-based ionosphere GNSS_VTEC;

[0007] Read the atmospheric density profile collected by the COSMIC satellite to obtain the VTEC observed by the COSMIC satellite, that is, COSMIC_VTEC, and integrate and save the GNSS_VTEC and COSMIC_VTEC data within the selected time period;

[0008] Through the integrated COSMIC_VTEC, register the GNSS_VTEC within the preset time range and the limited geographical location range, and input the registered data into the BP neural network for processing to output the VTEC prediction difference;

[0009] Correct the COSMIC_VTEC according to the VTEC prediction difference to obtain the corrected data.

[0010] Preferably, the carrier-smoothed pseudorange method is used to solve the slant total electron content STEC on the line of sight from the ground-based observation station to the satellite, specifically:

[0011]

[0012] In the formula, f1 and f2 are the carrier frequencies of the L1 band and the L2 band respectively; P sm is the smoothed pseudorange combined observation value; DCB i and DCB j are the hardware delay biases of the station and the satellite respectively; the unit of STEC is TECU, 1 TECU = 10 16 electrons / m 2 ; c is the hardware delay correction constant, which is used to represent the delay difference correction related to the frequency band between the receiver and the satellite.

[0013] Preferably,

[0014] VTEC = F (az) STEC;

[0015] In the formula, F (az) is an intermediate parameter, and the specific calculation is as follows:

[0016]

[0017] where R is the radius of the earth, H is the height of the ionosphere, and az is the zenith distance at the piercing point.

[0018] Preferably, the GNSS_VTEC and COSMIC_VTEC data within the selected time period are integrated and processed, including:

[0019] Adding characteristic variables to the ground-based ionospheric vertical total electron content GNSS_VTEC and the occultation ionospheric vertical total electron content COSMIC_VTEC according to the same time node, and unifying the data formats.

[0020] Preferably, the GNSS_VTEC within the preset time range and the limited geographical location range is registered, including:

[0021] Using the threshold matching method to match the ground-based GNSS stations that meet the preset time range and the limited geographical location range, calculating the weighted distance from the matched GNSS stations to the COSMIC occultation profile, and obtaining the registration result.

[0022] Preferably, the weighted distance from the matched GNSS station to the COSMIC occultation profile is calculated as:

[0023]

[0024] In the formula, ω i is the inverse distance weight, d i is the distance between the i-th GNSS station and the COSMIC occultation profile within the same COSMIC profile in the preset time range and the limited geographical location range, is the weighted distance after registration of the COSMIC occultation profile and the nearby GNSS stations.

[0025] Preferably, if the COSMIC occultation profile corresponds to multiple GNSS stations within the specified time and space range, the inverse distance weight method is used to solve the weighted distance between the GNSS_VTEC data after registration and the GNSS stations to the COSMIC occultation profile.

[0026] Preferably, the data after registration is input into a BP neural network for processing, including:

[0027] The registered data is input into the BP neural network, and the neurons are used to learn the error changes between the multi-source VTEC data, and finally the difference between COSMIC_VTEC and GNSS_VTEC is predicted, that is, the VTEC prediction difference.

[0028] Preferably, correcting the COSMIC_VTEC according to the VTEC prediction difference includes:

[0029] The VTEC prediction difference is added to the COSMIC_VTEC to obtain the corrected data, i.e. COSMIC_VTEC corr .

[0030] Preferably, the method further includes, if the corrected data does not achieve the expected effect, adjusting the number of hidden layers and the number of neurons of the BP neural network, thereby ultimately achieving high-precision COSMIC_VTEC data correction; wherein the correlation coefficient, mean absolute error and root mean square error are used as accuracy measurement indicators.

[0031] Compared with the prior art, the present invention has the following advantages and technical effects:

[0032] In order to solve the problem of low quality of ionospheric VTEC data observed by airborne COSMIC constellation, the present invention adopts VTEC data of ground-based GNSS as the benchmark, and corrects COSMIC_VTEC data by inverse distance weighting method and BP neural network. corr The data accuracy is comparable to that of ground-based GNSS, providing more reliable data support for space weather forecasting, navigation positioning, and related application areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0034] Figure 1 This is a flow chart of a multi-device system error correction method for ionospheric total electron content according to an embodiment of the present invention;

[0035] Figure 2 Schematic diagram of the distribution of multiple GNSS stations corresponding to the COSMIC occultation profile according to an embodiment of the present invention;

[0036] Figure 3The distribution diagram of the prediction effect of the BP neural network structure according to the embodiment of the present invention. Among them, the X-axis is the number of neurons in the first hidden layer of the BP neural network, the Y-axis is the number of neurons in the second hidden layer of the BP neural network, and the Z-axis is the number of neurons in the third hidden layer of the BP neural network;

[0037] Figure 4 The distribution diagram of the RMSE results of the COSMIC_VTEC data before and after calibration according to the embodiment of the present invention. Among them, (a) is the distribution diagram of the correlation error before calibration, and (b) is the distribution diagram of the correlation error after calibration;

[0038] Figure 5 The distribution diagram of the correlation error of the COSMIC_VTEC data in each season before and after calibration according to the embodiment of the present invention. Among them, (a), (c), (e), and (g) are all images before calibration, (b), (d), (f), and (h) are all images after calibration, (a) and (b) represent spring, (c) and (d) represent summer, (e) and (f) represent autumn, and (g) and (h) represent winter. Detailed implementation manners

[0039] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0040] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0041] As a complex atmosphere affected by multiple factors such as solar activity, geomagnetic activity, and climate change, the ionosphere has a high degree of dynamicity and uncertainty. The present invention adopts a calibration method between multi-source data, that is, taking the ground-based GNSS_VTEC data as the benchmark, and using a multi-layer backpropagation neural network to calibrate the observations of the space-based COSMIC to the benchmark level. This method can effectively and efficiently calibrate the observations of the space-based COSMIC, thereby providing rich and accurate data support for realizing high-precision and high spatio-temporal resolution ionospheric modeling.

[0042] Explanations for the technical terms in this embodiment are as follows:

[0043] Ground-based GNSS refers to the Global Navigation Satellite System (GNSS) receiving devices installed on the ground, which are used to receive satellite signals and perform positioning and navigation. The GNSS system includes satellite navigation systems of multiple countries, such as GPS in the United States, GLONASS in Russia, Beidou system in China, Galileo system in the European Union, etc. These systems provide positioning, navigation and time services globally through the radio signals transmitted by satellites.

[0044] Ground-based GNSS is widely used in various fields, including but not limited to:

[0045] Engineering surveying: In aspects such as building construction, topographic surveying, and deformation monitoring, ground-based GNSS can provide high-precision position information.

[0046] Agriculture: In precision agriculture, ground-based GNSS can help farmers with precise sowing, fertilization, and harvesting, improving agricultural production efficiency.

[0047] Transportation: In fields such as logistics, shipping, and aviation, ground-based GNSS is used for navigation and positioning to ensure safety and efficiency.

[0048] Emergency response: In natural disasters and emergencies, ground-based GNSS provides fast and accurate positioning information to help rescue personnel reach the scene quickly.

[0049] The working principle of ground-based GNSS is based on the basic principle of satellite navigation systems, that is, calculating the user's position by measuring the satellite signal propagation time. The specific steps include:

[0050] Signal reception: The receiver receives signals from multiple satellites.

[0051] Time calculation: By recording the signal propagation time, calculate the distance between the satellite and the receiver.

[0052] Position solution: Using the data of multiple satellites, calculate the specific position of the receiver through algorithms.

[0053] COSMIC (Constellation Observing System for Meteorology, Ionosphere and Climate) is a space-based GPS constellation observing system composed of 6 low-earth orbit satellites, mainly used for the observation of weather, climate and ionosphere.

[0054] The COSMIC system has been in operation since September 2006 and can provide approximately 2,000 to 3,000 occultation points of data globally every day. These data cover the atmospheric temperature, pressure, and humidity profile information from an altitude of 40 kilometers to near the ground. Through these data, the temperature, refractive index, and humidity profiles in the atmosphere can be retrieved, and the accuracy and usability of these data have been verified. In particular, the data in cloud areas have a high degree of agreement with radiosonde data, demonstrating the advantage of its all-weather performance.

[0055] The main functions of the COSMIC system include:

[0056] Weather and climate monitoring: Provide meteorological data globally to assist in predicting weather changes and climate changes.

[0057] Ionospheric observation: Monitor the changes in the ionosphere, which has an important impact on radio communication and navigation systems.

[0058] The COSMIC system has a wide range of application fields. It can not only be used for weather forecasting and climate research, but also improve the initial field of numerical weather prediction through the atmospheric data it provides, thereby enhancing the accuracy of weather prediction.

[0059] In this embodiment, a total of 184 GNSS stations in the mid-low latitude regions of the world are selected and evenly distributed in the research area; observational data and navigation files spanning more than three years from October 1, 2019 to December 31, 2023 are selected to ensure sufficient data volume and guarantee the effect of the subsequent correction model. Figure 2 It is a schematic diagram of the distribution of multiple GNSS stations corresponding to the COSMIC occultation profile.

[0060] A multi-device systematic error correction method for the total electron content in the ionosphere, as Figure 1 , specifically includes:

[0061] Obtain the dual-frequency observables collected by the GNSS ground-based observation stations in the research area, use the carrier-smoothed pseudorange method to solve the slant total electron content STEC on the line of sight from the ground-based observation station to the satellite, and use the mapping function to calculate and obtain the ground-based ionospheric vertical total electron content GNSS_VTEC;

[0062] Read the atmospheric density profile collected by the COSMIC satellite to obtain the VTEC observed by the COSMIC satellite, that is, COSMIC_VTEC, and integrate and process the GNSS_VTEC and COSMIC_VTEC data within the selected time period and save them;

[0063] By integrating the processed COSMIC_VTEC, aligning the GNSS_VTEC within a preset time range and a limited geographical location range, and inputting the aligned data into a BP neural network for processing to output a VTEC prediction difference;

[0064] The COSMIC_VTEC is corrected according to the VTEC prediction difference to obtain corrected data.

[0065] The present invention uses the VTEC data of ground-based GNSS as the benchmark, and corrects the COSMIC_VTEC data by using the inverse distance weighting method and BP neural network. corr The data accuracy is comparable to that of ground-based GNSS, providing more reliable data support for space weather forecasting, navigation positioning, and related application areas.

[0066] Furthermore, we obtained original observation files and satellite navigation messages from 184 ground-based GNSS stations and COSMIC occultation profiles around the world, and used a carrier phase smoothed pseudorange algorithm to calculate the slant total electron content (STEC) on the line of sight from the station to the satellite. We converted STEC into vertical total electron content (VTEC) using the Abel inversion technique, and then used a carrier phase smoothed pseudorange method to calculate the slant total electron content (STEC) on the line of sight from the ground-based station to the satellite. Specifically,

[0067]

[0068] Where f1 and f2 are the carrier frequencies of L1 and L2 bands respectively; P sm is the smoothed pseudorange combined observation value; DCB i and DCB j are the hardware delay deviations of the station and the satellite respectively; the unit of STEC is TECU, 1TECU=10 16 electrons / m 2 ; c is the hardware delay correction constant, which is usually a known value used to represent the frequency band-related delay difference correction between the receiver and the satellite.

[0069] Specifically:

[0070] VTEC=F (az) STEC;

[0071] Where, F (az) is an intermediate parameter, and the specific calculation is:

[0072]

[0073] Where R is the radius of the Earth, H is the height of the ionosphere, and az is the zenith distance at the puncture point.

[0074] Furthermore, the GNSS_VTEC and COSMIC_VTEC data within the selected time period are integrated and processed, including:

[0075] The ground-based ionospheric vertical total electron content GNSS_VTEC and COSMIC_VTEC are added with characteristic variables at the same time nodes, and the data format is unified.

[0076] Specifically, in this embodiment, the solved GNSS station VTEC data and COSMIC_VTEC are added with Dst, KP, and F10.7 as feature variables at the same time node, and the data format is saved in .txt format to facilitate data registration.

[0077] Furthermore, GNSS_VTEC within a preset time range and a limited geographical location range is aligned, including:

[0078] A threshold matching method is used to match ground-based GNSS stations that meet the preset time range and the limited geographical location range, and the weighted distance from the matched GNSS station to the COSMIC occultation profile is calculated to obtain the registration result.

[0079] Specifically, in this embodiment, a threshold matching method is used to match ground-based GNSS stations that meet the COSMIC occultation profile within a 10-minute time range and a geographical location range of 5 longitudes and 3 latitudes. The weighted distance from the matched GNSS station to the COSMIC occultation profile is calculated and input into the prediction model as a feature vector to improve the prediction accuracy.

[0080] The weighted distance from the GNSS station to the COSMIC occultation profile after calculation and matching is:

[0081]

[0082] Where, ω i is the inverse distance weight, d i is the distance between the i-th GNSS station and the COSMIC occultation profile line within the preset time range and limited geographical location on the same COSMIC profile line, is the distance between the COSMIC occultation profile and the nearby GNSS stations after registration.

[0083] The formula for calculating distance based on longitude and latitude is as follows:

[0084]

[0085] Where d is the distance between the GNSS station and the COSMIC occultation profile, R is the radius of the Earth, and are the latitudes of the GNSS station and the COSMIC occultation profile respectively, is the latitude difference between the GNSS station and the COSMIC occultation profile, and Δλ is the longitude difference between the GNSS station and the COSMIC occultation profile.

[0086] Furthermore, when the COSMIC occultation profile corresponds to multiple GNSS stations within a specified time and space range, the inverse distance weighting method is used to solve the registered GNSS_VTEC data and the weighted distance between the GNSS station and the COSMIC occultation, specifically:

[0087]

[0088] When matching, the distance between the GNSS station and the COSMIC occultation profile is generated according to the inverse distance weight ω i Based on the inverse distance weighting method, the reciprocal of the distance between the GNSS station and the COSMIC occultation profile is used as the weight, and d i is the distance between the i-th GNSS station and the COSMIC occultation profile within the time and longitude and latitude range specified by the same COSMIC profile. is the registered distance between the COSMIC occultation profile and the nearby GNSS station. In this embodiment, the direction is defined as positive north of the COSMIC profile and negative south of the profile. The side with the larger proportion of the inverse distance weight of the station is defined as the direction of the distance.

[0089]

[0090] Where GNSS_VTEC is the VTEC calculated using the inverse distance weighting method based on the GNSS station VTEC data within the time and space range specified by the COSMIC occultation profile. VTEC i is the VTEC of the i-th GNSS station.

[0091] Furthermore, the registered data is input into a BP neural network for processing, including:

[0092] The registered data is input into the BP neural network, and the neurons are used to learn the error change situation between the multi-source VTEC data, and finally the difference between the COSMIC_VTEC and the GNSS_VTEC is predicted, that is, the VTEC prediction difference.

[0093] Specifically, in this embodiment, a BP neural network is used for data prediction. The number of hidden layers of the BP neural network is set to 3 layers, and the prediction effect distribution is as follows Figure 3 shown. The BP neural network structure with the best prediction effect is (26, 24, 18), that is, the number of neurons in the first layer is 26, the number of neurons in the second layer is 24, and the number of neurons in the third layer is 18. The input data includes longitude and latitude, Dst, F10.7, KP, GNSS_VTEC, COSMIC_VTEC, time. Through the training of the BP neural network, the difference Δ (prediction difference of VTEC) between GNSS_VTEC and COSMIC_VTEC is predicted, and the predicted VTEC prediction difference is added to COSMIC_VTEC to obtain the corrected COSMIC_VTEC corr .

[0094] Specifically, a BP neural network is used for data prediction. The number of hidden layers of the neural network is set to 3 layers, and the prediction effect distribution is as follows Figure 3 shown. The BP neural network structure with the best prediction effect is (26, 24, 18), that is, the number of neurons in the first layer is 26, the number of neurons in the second layer is 24, and the number of neurons in the third layer is 18. The input data includes longitude and latitude, Dst, F10.7, KP, GNSS_VTEC, COSMIC_VTEC, time. Through the training of the BP neural network, the difference Δ between GNSS_VTEC and COSMIC_VTEC is predicted.

[0095] Furthermore, correcting COSMIC_VTEC according to the VTEC prediction difference includes:

[0096] Adding the VTEC prediction difference to the COSMIC_VTEC to obtain the corrected data, that is, COSMIC_VTEC corr .

[0097] Furthermore, the method further includes that if the corrected data does not meet the expected effect, the number of hidden layers and the number of neurons of the BP neural network are adjusted to finally achieve high-precision correction of COSMIC_VTEC data; among them, the correlation coefficient, mean absolute error, and root mean square error are used as precision measurement indicators.

[0098] Specifically, based on the GNSS_VTEC data as the true value to measure the corrected COSMIC_VTEC corrThe accuracy of the data. If the expected effect cannot be achieved, adjust the number of hidden layers and the number of neurons in the BP neural network to finally achieve high-precision COSMIC_VTEC data correction. The correlation coefficient (R), mean absolute error (MAE), and root mean square error (RMSE) are used as accuracy measurement indicators. The specific sizes of the indicators can be adjusted according to actual needs.

[0099]

[0100] In the formula, X represents the registered GNSS_VTEC, and Y represents the VTEC data of the COSMIC occultation profile (including the VTEC data of the COSMIC occultation profile before correction: COSMIC_VTEC, the VTEC data of the COSMIC occultation profile after correction: COSMIC_VTEC corr ) represents the mean of the registered GNSS_VTEC, represents the mean of the VTEC data of the COSMIC occultation profile (including the VTEC data of the COSMIC occultation profile before correction: COSMIC_VTEC, the VTEC data of the COSMIC occultation profile after correction: COSMIC_VTEC corr ) and n is the total amount of data after registration. Figure 4 (a)- Figure 4 (b) is the distribution map of the RMSE results of the COSMIC_VTEC data before and after correction. Figure 5 (a)- Figure 5 (h) are the distribution maps of the correlation errors of the COSMIC_VTEC data in each season before and after correction. Among them, (a), (c), (e), and (g) are all images before correction, (b), (d), (f), and (h) are all images after correction, (a) and (b) represent spring, (c) and (d) represent summer, (e) and (f) represent autumn, and (g) and (h) represent winter

[0101] MAE explains the average magnitude of the error and measures the closeness between the predicted value and the actual value without considering the direction of the error. R is a dimensionless value that represents the correlation strength between COSMIC_VTEC and GNSS_VTEC before and after correction. The value range of R is from -1 to 1. R approaching 1 indicates a strong positive linear relationship, approaching -1 indicates a strong negative linear relationship, and 0 indicates no linear relationship. RMSE reflects the average measure of the error magnitude. Due to squaring, larger errors have a greater impact on RMSE. Therefore, RMSE is more sensitive to outliers and is usually used to measure the accuracy of the prediction model.

[0102] In this embodiment, the observed values of space-based COSMIC are corrected by the vertical total electron content (VTEC) data of ground-based GNSS ionosphere, so as to eliminate the systematic errors between multiple devices and improve the accuracy and reliability of navigation and positioning.

[0103] The above are only the preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the technical field of the present application within the technical scope disclosed by the present application should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A multi-device systematic error correction method for the total electron content in the ionosphere, characterized in that include: The dual-frequency observations collected by the GNSS ground-based observation station in the study area were obtained. The carrier-smoothed pseudorange method was used to solve the slant total electron content (STEC) on the line of sight from the ground-based observation station to the satellite. The ground-based vertical total electron content (GNSS_VTEC) was obtained by using a mapping function. Read the atmospheric density profiles collected by the COSMIC satellite to obtain the VTEC observed by the COSMIC satellite, namely COSMIC_VTEC. Integrate and process the GNSS_VTEC and COSMIC_VTEC data within the selected time period and save them. By integrating the processed COSMIC_VTEC, aligning the GNSS_VTEC within a preset time range and a limited geographical location range, and inputting the aligned data into a BP neural network for processing to output a VTEC prediction difference; Correcting the COSMIC_VTEC according to the VTEC prediction difference to obtain corrected data; The integration and processing of GNSS_VTEC and COSMIC_VTEC data within the selected time period includes: Add characteristic variables to the ground-based ionospheric vertical total electron content GNSS_VTEC and the COSMIC_VTEC at the same time nodes and unify the data formats; Perform GNSS_VTEC registration within a preset time range and a limited geographic location, including: A threshold matching method is used to match ground-based GNSS stations that meet the preset time range and the limited geographical location range, and a weighted distance from the matched GNSS station to the COSMIC occultation profile is calculated to obtain a registration result; The formula for calculating distance based on longitude and latitude is as follows: ； Wherein, d is the distance between the GNSS station and the COSMIC occultation profile, R is the radius of the Earth, and are the latitudes of the GNSS station and the COSMIC occultation profile respectively, is the latitude difference between the GNSS station and the COSMIC occultation profile, is the longitude difference between the GNSS station and the COSMIC occultation profile; The weighted distance from the matched GNSS station to the COSMIC occultation profile is calculated as: ； ； Where, ω i is the inverse distance weight, d i is the distance between the i-th GNSS station and the COSMIC occultation profile line within the preset time range and limited geographical location on the same COSMIC profile line, is the weighted distance after registration between the COSMIC occultation profile and the nearby GNSS stations; If the COSMIC occultation profile corresponds to multiple GNSS stations within the specified time and space range, the inverse distance weighted method is used to solve the weighted distance between the registered GNSS_VTEC data and the GNSS station to the COSMIC occultation profile; The direction is defined as positive to the north of the COSMIC section line and negative to the south of the section line. The side with the larger inverse distance weight is defined as the distance direction; ； Wherein, GNSS_VTEC is the VTEC calculated by using the inverse distance weighting method based on the VTEC data of GNSS stations within the specified time and space range of the COSMIC occultation profile, and VTEC i is the VTEC of the i-th GNSS station; The registered data is input into the BP neural network for processing, including: The registered data is input into the BP neural network, and the neurons are used to learn the error changes between the multi-source VTEC data, and finally the difference between COSMIC_VTEC and GNSS_VTEC is predicted, that is, the VTEC prediction difference.

2. The method according to claim 1, characterized in that, The carrier-smoothed pseudorange method is used to solve the slant total electron content (STEC) on the line of sight from the ground-based observation station to the satellite, specifically: ； where f1 and f2 are the carrier frequencies of the L1 and L2 frequency bands respectively; P sm is the smoothed pseudorange combination observation; DCB i and DCB j are the hardware delay biases of the station and the satellite respectively; the unit of STEC is TECU, and 1 TECU = 10 16 electrons / m 2 ; c is the hardware delay correction constant, which is used to represent the delay difference correction related to the frequency band between the receiver and the satellite.

3. The method according to claim 2, wherein: ； where F (az) is an intermediate parameter, and the specific calculation is as follows: ； Where R is the radius of the Earth, H is the height of the ionosphere, and az is the zenith distance at the puncture point.

4. The method according to claim 1, wherein Correcting the COSMIC_VTEC according to the VTEC prediction difference includes: Add the VTEC prediction difference to the COSMIC_VTEC to obtain the corrected data, i.e., COSMIC_VTEC corr .

5. The method according to claim 1, wherein The method further includes that if the corrected data fails to achieve the expected effect, the number of hidden layers and the number of neurons of the BP neural network are adjusted, and finally, high-precision COSMIC_VTEC data correction is realized; among them, the correlation coefficient, the mean absolute error, and the root mean square error are used as precision measurement indicators.

Images