High-precision ionosphere three-dimensional electron density data reconstruction method and device

Abstract

The invention discloses a high-precision ionosphere three-dimensional electron density data reconstruction method and device, and the method comprises the steps: building a three-dimensional ionosphere electron density reconstruction model, taking actual observation data as the input data of the three-dimensional ionosphere electron density reconstruction model, and outputting the three-dimensional ionosphere electron density reconstruction data; establishing a TEC decomposition model, and partitioning the three-dimensional ionospheric electron density reconstruction data according to a space weather event to serve as training data of the TEC decomposition model; and after the TEC decomposition model is trained, inputting real-time two-dimensional TEC data and corresponding geological parameters into the TEC decomposition model to obtain high-precision three-dimensional ionized layer electron density reconstruction data. According to the method, subtle changes of the ionized layer can be captured more accurately, detailed three-dimensional electron density distribution is generated, the performance under high resolution is remarkably superior to that of a traditional technology, the reconstruction precision is improved, the adaptability and the real-time performance of the model are improved, and the accuracy and the reliability of data in the TEC decomposition process are ensured.

Description

Technical Field

[0001] The present invention belongs to the field of space model construction, and more specifically, relates to a method and device for reconstructing high-precision ionospheric three-dimensional electron density data. Background Art

[0002] Ionospheric electron density is a key indicator of ionospheric mobility and is crucial for applications such as shortwave technology, remote sensing systems, and satellite communications. Its accurate estimation ensures the stability and reliability of signal transmission and is of great significance for improving communication quality and system performance. At present, the following means are mainly used to monitor the global ionospheric electron density: (1) Global Positioning System (GPS): It can provide global data and is suitable for large-scale ionospheric monitoring. (2) Low Earth Orbit Satellite (LEO): COSMIC, Yunhai Satellite and other low-Earth orbit satellites provide global observations of ionospheric electron density by measuring the delay of the GPS signals they receive. (3) High Frequency Radar (HF Radar): HF radar is used to send signals to the ionosphere and infer the electron density by measuring the changes in the echo signals. (4) Global Ionospheric Model: The International Ionospheric Reference Model (IRI) and other models provide standardized reference data for global ionospheric electron density through data assimilation and model calculation.

[0003] However, due to the limitations of various models, the existing observation and reconstruction models cannot achieve accurate hierarchical reconstruction of the three-dimensional ionospheric electron density. Accurate reconstruction of the global ionospheric electron density profile faces the following problems. First, the observation data may be affected by instrument errors, spatial resolution limitations, environmental noise and signal interference, which in turn affects the data quality. Secondly, the global ionospheric observation data is unevenly distributed, especially in remote and high-latitude areas, resulting in data sparsity problems, which affects the accuracy of reconstruction. Thirdly, the accuracy of ionospheric models such as IRI is limited by the input data and model assumptions, which may cause the reconstruction results to be inconsistent with the actual situation. In addition, the computational complexity of processing large amounts of data and complex calculations, the spatial resolution limitations of existing observation techniques and models, and the dynamic changes of the global ionosphere will also affect the real-time and accuracy of the reconstruction.

[0004] The existing ionospheric electron density observation and reconstruction model has the following problems:

[0005] (1) Data sparsity: Existing ionospheric observation data are unevenly distributed around the world, especially in remote areas and high-latitude regions, which leads to data sparsity problems and affects the accuracy of precise reconstruction.

[0006] (2) Model accuracy limitation: The accuracy of existing ionospheric models (such as IRI) is limited by the input data and model assumptions, resulting in the reconstruction results of the three-dimensional electron density not being completely consistent with the actual situation.

[0007] (3) Computational complexity and real-time performance: Existing technologies may face the problem of insufficient computing resources when processing large amounts of data and complex calculations, which will limit the computing speed, affect real-time performance, and reduce the accuracy of the reconstruction results.

[0008] (4) Spatial resolution limitation: The spatial resolution of existing observation techniques and models is insufficient to effectively capture subtle changes in the ionosphere, which affects the accuracy of ionosphere reconstruction.

[0009] In short, the existing observation and reconstruction models cannot achieve accurate hierarchical reconstruction of the three-dimensional ionospheric electron density.

[0010] In view of this, overcoming the technical defects of the above-mentioned prior art is an urgent problem to be solved in this technical field. Summary of the invention

[0011] In view of the above defects or improvement needs of the prior art, the present invention provides a method and device for reconstructing high-precision three-dimensional ionospheric electron density data, the purpose of which is to achieve real-time and fast high-precision three-dimensional ionospheric electron density reconstruction by solving the inverse problem based on basic data such as the total electron content (TEC) of the ionosphere.

[0012] To achieve the above object, according to one aspect of the present invention, a method for reconstructing high-precision ionospheric three-dimensional electron density data is provided, the method comprising:

[0013] Establishing a three-dimensional ionospheric electron density reconstruction model, using actual observation data as input data of the three-dimensional ionospheric electron density reconstruction model, and outputting three-dimensional ionospheric electron density reconstruction data;

[0014] Establishing a TEC decomposition model, dividing the three-dimensional ionospheric electron density reconstruction data into blocks according to space weather events as training data for the TEC decomposition model;

[0015] After the TEC decomposition model is trained, real-time two-dimensional TEC data and corresponding geological parameters are input into the TEC decomposition model to obtain high-precision three-dimensional ionospheric electron density reconstruction data.

[0016] As a further improvement and supplement to the above solution, the present invention also includes the following additional technical features.

[0017] Preferably, the actual observation data includes time, longitude and latitude, solar zenith angle, altitude and geomagnetic activity index.

[0018] Preferably, after the three-dimensional ionospheric electron density reconstruction model is trained and evaluated, the accuracy of the three-dimensional ionospheric electron density reconstruction model is compared with the actual Cosmic satellite orbit observation data.

[0019] Preferably, after the accuracy of the three-dimensional ionospheric electron density reconstruction model reaches a preset threshold, the three-dimensional ionospheric electron density reconstruction data is divided into blocks according to space weather events as training data for the TEC decomposition model.

[0020] Preferably, the real-time two-dimensional TEC data in the TEC decomposition model and the three-dimensional ionospheric electron density reconstruction data maintain the same longitude and latitude resolution.

[0021] Preferably, the real-time two-dimensional TEC data in the TEC decomposition model includes time, TEC, geomagnetic activity index, longitude and latitude, and altitude.

[0022] Preferably, the geomagnetic activity index is time matched with real-time two-dimensional TEC data.

[0023] Preferably, the space weather events include geomagnetic storm periods and calm periods, each block contains a complete geomagnetic storm period or a complete calm period, and the order between each block is randomly disrupted when training the TEC decomposition model.

[0024] Preferably, the geological parameters include geomagnetic index, altitude and longitude and latitude data.

[0025] According to another aspect of the present invention, there is provided a device, the device comprising:

[0026] one or more processors;

[0027] A storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the high-precision ionospheric three-dimensional electron density data reconstruction method as described in the first aspect.

[0028] In general, the above technical solution conceived by the present invention has the following beneficial effects compared with the prior art:

[0029] 1. Event-specific learning: Data is divided into blocks according to geomagnetic storm events, so that the model can focus on learning the characteristics and laws during geomagnetic storms, thereby improving the model's reconstruction effect when similar events occur. The model can capture the specific impact pattern of geomagnetic storms on electron density and improve its response to the event.

[0030] 2. Improved generalization ability: By including data from geomagnetic storms and calm periods in the training data, the model can be trained under a variety of space weather conditions. This diverse training process enables the model to have stronger generalization ability when facing unseen events, that is, it can handle different types of ionospheric changes more accurately.

[0031] 3. Optimize the reconstruction effect: Focusing on the data blocks of geomagnetic storm events helps the model more accurately simulate and predict the behavior of the ionosphere under extreme conditions, thereby providing more accurate electron density reconstruction during geomagnetic storms. This precise reconstruction is of great significance for the optimization of space weather forecasting, navigation and communication systems.

[0032] 4. Maximizing data utilization: Time-block processing can make more efficient use of data resources, improve the performance of the overall model through the reconstruction of specific events, and reduce the problem of model overfitting due to data sparsity or specific conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0034] Figure 1 This is a schematic flow chart of a method for reconstructing high-precision ionospheric three-dimensional electron density data provided in the first embodiment;

[0035] Figure 2 This is a schematic diagram of a method for reconstructing high-precision ionospheric three-dimensional electron density data provided in the first embodiment;

[0036] Figure 3 This is a schematic diagram of a high-precision ionospheric three-dimensional electron density data reconstruction device provided in the second embodiment. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0038] Embodiment 1

[0039] Total electron content (TEC) is a key indicator for studying the Earth's ionosphere. TEC represents the total number of electrons that vertically pass through a unit area in the Earth's ionosphere, and the unit is usually the number of electrons or the amount of charge. The ionospheric electron density is the number of electrons present per unit volume, and there is a direct relationship between TEC and the ionospheric electron density. TEC can be calculated by integrating the ionospheric electron density over a vertical path. The ionosphere is usually divided into several layers, and the electron density is calculated in each layer, and then the electron density of these layers is integrated to obtain the value of TEC. Accordingly, the ionospheric electron density can be obtained by solving the inverse problem. TEC is usually measured by satellite systems such as the Global Positioning System (GPS), which can provide high-precision signal transmission data. Based on high-precision ionospheric TEC data, solving the inverse problem of ionospheric electron density can achieve accurate reconstruction of the electron density of different layers of the global ionosphere.

[0040] In recent years, deep learning technology has been widely used in the fields of nonlinear fitting and inverse problem modeling. The inverse problem solving method of ionospheric electron density based on deep learning can not only improve the accuracy of reconstruction, but also better capture the complex changes of the ionosphere, provide more detailed and reliable global ionospheric electron density distribution information, and provide strong support for ionospheric research and related applications.

[0041] Therefore, this embodiment proposes an inverse problem solving technology for solving the ionospheric electron density based on deep learning using the global ionospheric two-dimensional TEC data to solve the three-dimensional electron density. This embodiment uses two deep learning models and only based on basic data such as the ionospheric total electron content (TEC), by solving the inverse problem, it realizes real-time and fast high-precision three-dimensional ionospheric electron density reconstruction.

[0042] The first deep learning model is a three-dimensional ionospheric electron density reconstruction model. Based on the massive electron density data from the low-Earth orbit satellite (COSMIC), it uses time, solar zenith angle, longitude and latitude, altitude, and geomagnetic activity index data as constraints to reconstruct the three-dimensional electron density and provide training data for subsequent models.

[0043] The second deep learning model is a two-dimensional TEC decomposition model. Based on high-precision ionospheric TEC data such as IGS-TEC and MIT-TEC, the TEC data is decomposed into high-precision three-dimensional ionospheric electron density data with time, longitude and latitude, altitude, and geomagnetic activity index as constraints. After the model training is completed, the present invention can achieve real-time and accurate reconstruction of the global ionospheric three-dimensional electron density only through the two-dimensional TEC decomposition model and high-precision TEC and other basic data.

[0044] In the first embodiment, a method for reconstructing high-precision ionospheric three-dimensional electron density data is provided, and the method comprises the following steps: Figure 1 As shown:

[0045] S101: Establishing a three-dimensional ionospheric electron density reconstruction model, using actual observation data as input data of the three-dimensional ionospheric electron density reconstruction model, and outputting three-dimensional ionospheric electron density reconstruction data.

[0046] The actual observation data include time, longitude and latitude, solar zenith angle, altitude and geomagnetic activity index. In the process of model training, if all observation data are used as input and electron density is used as output for training, then the reconstruction stage must rely on all these input data for complete reconstruction. If only part of the data is used, it will lead to the loss of specific parameter constraints, thereby reducing the reconstruction accuracy of the model and directly affecting the performance and accuracy of the TEC decomposition model.

[0047] S102: Establishing a TEC decomposition model, dividing the three-dimensional ionospheric electron density reconstruction data into blocks according to space weather events as training data for the TEC decomposition model.

[0048] S103: After the TEC decomposition model is trained, real-time two-dimensional TEC data and corresponding geological parameters are input into the TEC decomposition model to obtain high-precision three-dimensional ionospheric electron density reconstruction data.

[0049] The high-precision ionospheric three-dimensional electron density data reconstruction method provided in this embodiment generates training data through a deep learning model: through deep learning technology, the model can more accurately capture subtle changes in the ionosphere and generate a detailed three-dimensional electron density distribution. The optimized training process, enhanced data processing algorithm, and real-time update capability make this method significantly better than traditional technologies at high resolution. This innovative method not only improves the reconstruction accuracy, but also improves the adaptability and real-time performance of the model, ensuring the accuracy and reliability of the data during the TEC decomposition process.

[0050] The high-precision ionospheric three-dimensional electron density data reconstruction method provided in this embodiment optimizes real-time prediction capability and computational efficiency: it can predict ionospheric electron density in real time, reduce computational complexity and resource consumption, improve processing speed, and support real-time application requirements.

[0051] The high-precision ionospheric three-dimensional electron density data reconstruction method provided in this embodiment reduces the impact of data sparsity: by utilizing extensive TEC observation data and deep learning technology, it reduces the dependence on traditional data sparsity and improves the coverage and accuracy of ionospheric electron density reconstruction on a global scale.

[0052] The high-precision ionospheric three-dimensional electron density data reconstruction method provided in this embodiment simplifies the data processing process: in the application process, the trained model and high-precision TEC data can be directly used to achieve accurate reconstruction of the ionospheric three-dimensional electron density, thereby simplifying the data processing process.

[0053] In combination with the embodiment of the present invention, there is also a preferred implementation scheme, specifically, the actual observation data includes time, longitude and latitude, solar zenith angle, altitude and geomagnetic activity index.

[0054] The change of ionospheric electron density is closely related to time, spatial position, solar activity and geomagnetic activity. Therefore, the present invention selects time, longitude and latitude, solar zenith angle (SZA), altitude, and geomagnetic index data (AE, SYM-H, F10.7) as input data of the three-dimensional electron density reconstruction model. Among them, time, longitude and latitude, altitude, and electron density data are derived from COSMIC occultation data, and geomagnetic index data are derived from the omni data set. The solar zenith angle data is based on the electron density data provided by the ionospheric COSMIC satellite, and is matched and calculated by the time corresponding to the data. First, the mean solar time angle (Mean Solar Time Angle) of the sun is calculated based on the annual accumulation day and observation time, and then the equal time difference (Equation of Time) and the solar declination angle (Solar Declination) are derived. Next, the time offset related to the solar time angle is calculated to determine the true solar time (True Solar Time), and the solar time angle (Hour Angle) is derived therefrom. Finally, the relevant angle is converted into radians, and the solar zenith angle (SZA) is accurately calculated by the trigonometric formula. This calculation process fully considers the influence of astronomical factors such as the Earth's rotation, the eccentricity of the Earth's orbit, and the obliquity of the ecliptic, ensuring the high accuracy of the SZA calculation. The solar zenith angle (SZA) is the angle between the sun's rays and the zenith direction of the observation point. The change in the ionospheric electron density is closely related to solar activity. The change in the solar zenith angle is affected by the season and latitude. The change pattern in different seasons and latitudes will affect the seasonal and diurnal variation characteristics of the ionospheric electron density. Calculating the impact of the solar zenith angle (SZA) on the ionospheric electron density can significantly improve the accuracy of the ionospheric model and improve the accuracy of the electron density.

[0055] In combination with the embodiment of the present invention, there is also a preferred implementation scheme. Specifically, after the three-dimensional ionospheric electron density reconstruction model is trained and evaluated, the accuracy of the three-dimensional ionospheric electron density reconstruction data output is tested and compared.

[0056] In the model evaluation process, the ionospheric electron density data generated by the reconstructed model must first be compared with the actual observation data of the Cosmic satellite orbit. This comparison can verify the accuracy of the model in space and time. At the same time, the electron density data generated by the International Ionospheric Reference Model (IRI) must also be compared with the actual observation data of the satellite orbit to evaluate the benchmark performance of the IRI model. Through this double comparison, the degree of improvement of the reconstructed model can be evaluated more objectively. IRI is a commonly used model. By evaluating the performance of IRI and then comparing it with the performance of the reconstructed model, the performance of the reconstructed model can be more objectively explained.

[0057] The calculation of mean relative error (MRE) and root mean square error (RMSE) is to quantify the difference between the model output and the actual observation data. The calculation of MRE and RMSE is based on a point-by-point comparison of the ionospheric electron density generated by the model and the electron density data observed by the Cosmic satellite. MRE reflects the relative deviation of the model relative to the observed value, while RMSE quantifies the overall size of the model prediction error. When MRE and RMSE reach the preset threshold, it can be judged that the accuracy of the ionospheric electron density generated by the three-dimensional ionospheric electron density reconstruction model reaches the preset threshold.

[0058] In the first embodiment, the real-time high-resolution two-dimensional TEC data and three-dimensional electron density data are assimilated together through the TEC decomposition model, and the data advantages of TEC are applied to the three-dimensional electron density data in the form of neural network modeling. First, the architecture of the deep neural network (DNN) is defined, including an input layer, multiple hidden layers, and an output layer. The input layer receives time, TEC (total electron content), geomagnetic activity index, longitude and latitude, and altitude as features, and the output layer predicts electron density data. The ReLU activation function is selected to introduce nonlinear characteristics. During the training process, the mean square error (MSE) is used as the loss function, and the Adam optimizer is used to optimize the model parameters. The data used for training the TEC decomposition model is a total of 18.8 billion data from 2011 to 2017, and the data from 2018 to 2019 is used as a test set. The network is optimized by forward propagation and backpropagation algorithms, and the model performance is evaluated using the validation set after each training cycle to adjust the hyperparameters to improve the generalization ability of the model. After training, the root mean square error (RMSE) and mean absolute error (MAE) are calculated using the test set data to evaluate the prediction accuracy and explanatory power of the TEC decomposition model. The calculation of MRE and RMSE is based on a point-by-point comparison of the ionospheric electron density generated by the model and the electron density data observed by the Cosmic satellite, analyzing the deviation between the prediction results of the TEC decomposition model and the actual data, and checking the stability of the model. According to the test results, necessary model optimization is performed, and the network architecture or hyperparameters are adjusted to improve the final performance.

[0059] In combination with the embodiment of the present invention, there is also a preferred implementation scheme. Specifically, after the accuracy of the three-dimensional ionospheric electron density reconstruction model reaches a preset threshold, the three-dimensional ionospheric electron density reconstruction data is divided into blocks according to space weather events as training data for the TEC decomposition model.

[0060] In this embodiment, in the data preparation stage, the time resolution of the ionospheric global electron density data is 1h, the altitude resolution is 10km, the longitude resolution is 5.625°, and the latitude resolution is 2.8125°. The IGS-TEC data is linearly interpolated and reconstructed to maintain the same longitude and latitude resolution. During model training, it is necessary to compare the electron density output by the decomposition model with the high-precision electron density data generated by the reconstruction model. Therefore, it is necessary to adjust the time and longitude resolution of the TEC data input by the decomposition model to ensure that the output electron density matches the resolution of the high-precision electron density data generated by the reconstruction model.

[0061] In order to effectively model and predict the ionospheric three-dimensional electron density data during geomagnetic storm events, the training set is divided according to space weather events (geomagnetic storm period and calm period) in this embodiment 1 to ensure that the model can learn for different event states. Specifically, the data set is divided into several data blocks, each block represents a time period, such as 6 days of data, in order to capture a complete geomagnetic storm event, every 6 days of data is divided into a block, the order of data in each block remains unchanged, and the order between blocks is randomly shuffled.

[0062] In order to avoid bias in model training due to the order of data blocks, these blocks are shuffled during the training process. This shuffling strategy ensures that during the mini-batch stochastic gradient descent (SGD) process, the model is exposed to data samples from geomagnetic storms and non-geomagnetic storms, so that each training batch includes data from different event states. This method allows the model to learn the characteristics of geomagnetic storms and calm periods separately during training, thereby improving the model's generalization ability for data under different space weather conditions. Through this event-driven data partitioning and inter-block shuffling mechanism, the model can better cope with changes in data distribution in practical applications, improving prediction accuracy and adaptability.

[0063] In combination with the embodiment of the present invention, there is also a preferred implementation scheme, specifically, the real-time two-dimensional TEC data in the TEC decomposition model and the three-dimensional ionospheric electron density reconstruction data maintain the same longitude and latitude resolution.

[0064] The input of the TEC decomposition model is TEC data, and the output is electron density data, which realizes the mapping of two-dimensional TEC data to three-dimensional electron density data. The two data features are fused together during the training process of the TEC decomposition model.

[0065] In combination with the embodiment of the present invention, there is also a preferred implementation scheme. Specifically, the real-time two-dimensional TEC data in the TEC decomposition model includes time, TEC, geomagnetic activity index, longitude and latitude, and altitude.

[0066] In the process of training the TEC decomposition model, the consistency of data such as the geomagnetic index, altitude, longitude and latitude must be ensured to ensure the coordination between the data, thereby providing accurate and complete input for the model and ensuring the correct training and performance optimization of the TEC decomposition model.

[0067] This embodiment does not require the reconstructed ionospheric electron density profile data as a basis. This embodiment only uses a trained TEC decomposition model, high-precision TEC data as the main input, and geomagnetic index, altitude, longitude and latitude data as auxiliary inputs to achieve reconstruction of the global ionospheric three-dimensional electron density data.

[0068] TEC data provides obvious advantages over electron density data due to its direct measurement, high spatial resolution, real-time update and sensitivity to space weather events. This embodiment assimilates real-time and high-resolution two-dimensional TEC data and three-dimensional electron density data through a decomposition model, and uses the real-time advantage of TEC data to achieve decomposition modeling of electron density data with a neural network.

[0069] In combination with the embodiment of the present invention, there is also a preferred implementation scheme, specifically, the geomagnetic activity index is time matched with the real-time two-dimensional TEC data.

[0070] In combination with the embodiment of the present invention, there is also a preferred implementation scheme, specifically, the space weather event includes geomagnetic storm periods and calm periods, each block contains a complete geomagnetic storm period or a complete calm period, and the order between each block is randomly disrupted when training the TEC decomposition model.

[0071] The time of the 2D TEC data needs to correspond to the geomagnetic activity index of the 2D TEC data.

[0072] In combination with the embodiment of the present invention, there is also a preferred implementation scheme, specifically, the geological parameters include geomagnetic index, altitude and longitude and latitude data.

[0073] The geomagnetic index reflects the degree to which the Earth's magnetosphere is disturbed by the solar wind, and can significantly affect the state of the ionosphere, especially during magnetic storms. Therefore, the geomagnetic index is indispensable for accurately capturing the changes in the ionospheric electron density under different space weather conditions.

[0074] The electron density has significant variation characteristics at different altitudes, especially in the E layer and F layer of the ionosphere. Altitude is an important one-dimensional data in three-dimensional reconstruction, which determines that the TEC decomposition model can distinguish the electron density distribution at different altitudes. Without altitude data, the TEC decomposition model cannot generate a vertical profile of the ionosphere.

[0075] The electron density distribution in the ionosphere has obvious spatial dependence and is closely related to the geographical location. The longitude and latitude data are the basis for three-dimensional spatial positioning and are used to determine the electron density distribution in different areas on the earth's surface.

[0076] As long as the TEC data of the corresponding longitude and latitude are given, the data of every altitude in the ionosphere can be reconstructed through the model.

[0077] Method of using two-dimensional TEC to solve three-dimensional electron density: Compared with electron density data, TEC data provides obvious advantages over electron density data due to its direct measurement, high spatial resolution, real-time update and sensitivity to space weather events. Based on the deep learning method, two-dimensional TEC is used to solve three-dimensional electron density, and the data advantages of TEC are applied to three-dimensional electron density data in the form of neural network modeling.

[0078] Data processing mode for training data of two-dimensional TEC decomposition model: The training data of two-dimensional TEC decomposition model comes from the three-dimensional ionospheric electron density reconstruction model based on deep learning. The high-precision electron density data generated by the reconstruction model and the high-precision TEC data are matched with each other as the training data of the decomposition model. In the training process of the decomposition model, the global electron density profile data of the ionosphere generated every hour is matched with the real observed total electron content (TEC) data through precise time matching.

[0079] Block training method of two-dimensional TEC decomposition model: The event-based training method adopts the method of dividing the training data into blocks in accordance with space weather events (geomagnetic storm periods and calm periods), which is conducive to the fitting of the TEC decomposition model to different events and improves the generalization ability of the TEC decomposition model for specific space weather events.

[0080] When processing ionospheric electron density data, space weather events such as geomagnetic storms have a significant impact on data reconstruction. During geomagnetic storms, the ionospheric electron density undergoes drastic changes, which makes the reconstruction of electron density data during this period particularly important. By chunking the data by time, especially by dividing the data into periods of geomagnetic storm events and calm periods, the deep learning TEC decomposition model can be more focused on these specific events, thereby improving the reconstruction accuracy and the generalization ability of the TEC decomposition model. Specifically, by dividing the time periods by "events" during data processing and training, the TEC decomposition model is able to specifically learn the characteristics of electron density changes under different space weather conditions.

[0081] The global ionosphere three-dimensional electron density is reconstructed through neural network technology, which solves the problem of discrete observation data. The model generates high-precision three-dimensional electron density data on a global scale.

[0082] Neural network technology can effectively capture the complex relationship between ionospheric electron density and its influencing factors through its deep network structure and complex nonlinear activation function. Multi-level feature extraction and adaptive learning capabilities enable the TEC decomposition model to improve prediction accuracy and model ionospheric changes more accurately.

[0083] Neural network technology uses parallel computing and graphics processing unit (GPU) acceleration technology to greatly improve computing efficiency. By optimizing algorithms (such as Adam) and efficient network architecture, it can reduce computational complexity and increase computing speed when processing large-scale data.

[0084] During the training process, high-precision ionospheric total electron content (TEC) data are used to generate a high-precision three-dimensional electron density reconstruction model. The two-dimensional TEC decomposition model uses the direct measurement, high spatial resolution and real-time update characteristics of TEC data to carefully reconstruct the electron density through high spatial resolution TEC data. The real-time update capability of TEC data enables the model to reflect the latest state of the ionosphere in a timely manner, while the sensitivity to space weather events ensures that the model can provide a more accurate electron density distribution when events such as geomagnetic storms occur.

[0085] The reconstruction method of high-precision ionospheric three-dimensional electron density data not only improves the accuracy of the reconstruction results, but also enhances the adaptability of the model under different space weather conditions, and realizes real-time decomposition of three-dimensional electron density with high spatial resolution.

[0086] Embodiment 2:

[0087] A high-precision reconstruction device for ionospheric three-dimensional electron density data, such as Figure 3 As shown, the device includes:

[0088] one or more processors;

[0089] A storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement a method for reconstructing high-precision ionospheric three-dimensional electron density data as described in any one of the first embodiments.

[0090] Figure 3 A schematic diagram of the structure of a device for reconstructing high-precision ionospheric three-dimensional electron density data provided in the second embodiment. Figure 3 A block diagram of an exemplary high-precision ionospheric three-dimensional electron density data reconstruction device suitable for implementing an embodiment of the present invention is shown. Figure 3 The high-precision ionospheric three-dimensional electron density data reconstruction device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0091] like Figure 3 As shown, the high-precision ionospheric three-dimensional electron density data reconstruction device is presented in the form of a general device. The components of the high-precision ionospheric three-dimensional electron density data reconstruction device may include but are not limited to: one or more processors or processing units, memory, and a bus connecting different system components (including memory and processing unit).

[0092] The term "bus" refers to one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0093] The high-precision ionospheric three-dimensional electron density data reconstruction device typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the device that corrects the intelligent logging interpretation model, including volatile and non-volatile media, removable and non-removable media.

[0094] The memory may include a computer system readable medium in the form of a volatile memory, such as a random access memory (RAM) 30 and / or a cache memory. The high-precision ionospheric three-dimensional electron density data reconstruction device may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system may be used to read and write non-removable, non-volatile magnetic media ( Figure 3 not shown, usually called a "hard drive"). Although Figure 3 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical medium) may be provided. In these cases, each drive may be connected to the bus via one or more data medium interfaces. The memory may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the embodiments of the present invention.

[0095] A program / utility having a set (at least one) of program modules may be stored, for example, in a memory, such program modules including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules generally perform the functions and / or methods of the embodiments described herein.

[0096] The device for reconstructing high-precision ionospheric three-dimensional electron density data can also communicate with one or more external devices (such as keyboards, pointing devices, displays, etc.), and can also communicate with one or more devices that enable users to interact with the device for reconstructing high-precision ionospheric three-dimensional electron density data, and / or communicate with any device that enables the device for reconstructing high-precision ionospheric three-dimensional electron density data to communicate with one or more other devices (such as network cards, modems, etc.). Such communication can be carried out through an input / output (I / O) interface. In addition, the device for correcting the intelligent logging interpretation model can also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN) and / or a public network, such as the Internet) through a network adapter. Figure 3 As shown, the network adapter communicates with other modules of the high-precision ionospheric three-dimensional electron density data reconstruction device through the bus. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the high-precision ionospheric three-dimensional electron density data reconstruction device, including but not limited to: microcode, device driver, redundant processing unit, external disk drive array, RAID system, tape drive and data backup storage system, etc.

[0097] The processing unit executes various functional applications and data processing by running the program stored in the memory, for example, realizing the high-precision ionospheric three-dimensional electron density data reconstruction method provided by any embodiment of the present invention. That is: establishing a three-dimensional ionospheric electron density reconstruction model, using actual observation data as input data of the three-dimensional ionospheric electron density reconstruction model, and outputting three-dimensional ionospheric electron density reconstruction data; establishing a TEC decomposition model, dividing the three-dimensional ionospheric electron density reconstruction data into blocks according to space weather events as training data of the TEC decomposition model; after the TEC decomposition model is trained, inputting real-time two-dimensional TEC data and corresponding geological parameters into the TEC decomposition model to obtain high-precision three-dimensional ionospheric electron density reconstruction data.

[0098] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for reconstructing high-precision ionospheric three-dimensional electron density data, characterized in that: The method comprises: Establishing a three-dimensional ionospheric electron density reconstruction model, using actual observation data as input data of the three-dimensional ionospheric electron density reconstruction model, and outputting three-dimensional ionospheric electron density reconstruction data; Establishing a TEC decomposition model, dividing the three-dimensional ionospheric electron density reconstruction data into blocks according to space weather events as training data for the TEC decomposition model; After the TEC decomposition model is trained, real-time two-dimensional TEC data and corresponding geological parameters are input into the TEC decomposition model to obtain high-precision three-dimensional ionospheric electron density reconstruction data.

2. The method for reconstructing high-precision ionospheric three-dimensional electron density data according to claim 1, characterized in that: The actual observation data include time, longitude and latitude, solar zenith angle, altitude and geomagnetic activity index.

3. The method for reconstructing high-precision ionospheric three-dimensional electron density data according to claim 1, characterized in that: After training and evaluation, the three-dimensional ionospheric electron density reconstruction model is compared with the actual observation data of the Cosmic satellite orbit to determine the accuracy of the three-dimensional ionospheric electron density reconstruction model.

4. The method for reconstructing high-precision ionospheric three-dimensional electron density data according to claim 3, characterized in that: After the accuracy of the three-dimensional ionospheric electron density reconstruction model reaches a preset threshold, the three-dimensional ionospheric electron density reconstruction data is divided into blocks according to space weather events as training data for the TEC decomposition model.

5. The method for reconstructing high-precision ionospheric three-dimensional electron density data according to claim 1, characterized in that: The real-time two-dimensional TEC data in the TEC decomposition model and the three-dimensional ionospheric electron density reconstruction data maintain the same longitude and latitude resolution.

6. The method for reconstructing high-precision ionospheric three-dimensional electron density data according to claim 1, characterized in that: The real-time two-dimensional TEC data in the TEC decomposition model includes time, TEC, geomagnetic activity index, longitude and latitude, and altitude.

7. The method for reconstructing high-precision ionospheric three-dimensional electron density data according to claim 6, characterized in that: The geomagnetic activity index is time matched to real-time two-dimensional TEC data.

8. The method for reconstructing high-precision ionospheric three-dimensional electron density data according to claim 1, characterized in that: The space weather events include geomagnetic storm periods and calm periods, each block contains a complete geomagnetic storm period or a complete calm period, and the order between each block is randomly disrupted when training the TEC decomposition model.

9. The method for reconstructing high-precision ionospheric three-dimensional electron density data according to claim 1, characterized in that: The geological parameters include geomagnetic index, altitude and longitude and latitude data.

10. A high-precision ionospheric three-dimensional electron density data reconstruction device, characterized in that the device include: one or more processors; A storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the high-precision ionospheric three-dimensional electron density data reconstruction method as described in any one of claims 1-9.

Images