Low-latitude ionosphere disturbance and non-uniform body multi-means cooperative detection and classification method

Abstract

The invention relates to the technical field of satellite navigation and ionosphere detection, in particular to a low-latitude ionosphere disturbance and non-uniform body multi-means collaborative detection and classification method. A multi-network cooperation-multi-source fusion observation mode is adopted, full space-time coverage of low-latitude ionosphere disturbance is achieved, in addition, a data fusion method based on physical constraints is applied, an ionosphere-middle atmosphere-troposphere multi-physics field coupling constraint mechanism is constructed, space-time consistency of data obtained through different observation means is optimized, and the low-latitude ionosphere disturbance is obtained. And finally, key characteristic parameters are extracted by using a multi-scale signal processing method, a disturbance characteristic matrix is constructed in combination with radar observation data (such as drift speed and spectral width), a'three-axis nine-element 'disturbance / non-uniform body classification standard is proposed firstly, then deep learning classification is used, accurate classification of low-latitude ionosphere disturbance is realized, and GNSS signal anomaly recognition capability is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of satellite navigation and ionospheric detection, and in particular to a multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities. Background Art

[0002] The low-latitude ionosphere, influenced by solar radiation, dynamic processes, and geomagnetic activity, exhibits significant spatiotemporal heterogeneity. This is particularly true in the equatorial region, where complex structures such as proton plasma bubbles (EPBs), traveling ionospheric disturbances (TIDs), and ionospheric scintillations are prone to form. These anomalies can affect electromagnetic wave propagation, thereby reducing the accuracy and stability of the BeiDou / GNSS global navigation satellite system and other space communication systems. However, existing ionospheric detection methods suffer from problems such as limited spatiotemporal coverage, poor continuity of observation scales, poor consistency in the physical interpretation of observational data, and low accuracy in classifying disturbances / inhomogeneities. For example, GNSS TEC can provide large-scale information on the total electron density content, but has difficulty resolving fine vertical structures in the ionosphere. Coherent scatter radar can detect localized plasma inhomogeneities, but its spatial coverage is limited. Ionospheric altimeters can obtain electron density profiles, but due to the sparse distribution of observing stations, high-precision monitoring of regional ionospheric disturbances is difficult. In addition, in terms of anomaly classification, existing methods mainly rely on a single data source for classification and have no multi-physical field constraints, such as empirical classification based on the GNSS signal scintillation index (S4) or the TEC disturbance rate (ROTI). However, these methods have limited accuracy when faced with the complex disturbance environment of the low-latitude ionosphere, and it is difficult to fully characterize the formation mechanism of different types of disturbances. Summary of the Invention

[0003] The present invention aims to provide a multi-method collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities, overcoming the technical bottlenecks of existing methods that rely on a single data source, have limited observation coverage, and lack classification accuracy. This method integrates multiple detection methods such as GNSS TEC / scintillation monitors, ionospheric altimeters, and multi-radar systems in low-latitude regions through the construction of a "multi-network collaboration-multi-source fusion" observation mode. This method enables all-weather, cross-scale detection of low-latitude ionospheric disturbances, improving the spatiotemporal coverage and observation scale continuity of ionospheric detection. To enhance the ability to explain the physical driving mechanism, a "multi-physics field coupling constraint model" is proposed that integrates disturbance sources such as solar activity, geomagnetic disturbances, neutral wind fields, and gravity wave propagation. This model constructs a vertical coupling system of "ionosphere-middle atmosphere-troposphere" to enhance the consistency of physical interpretation of data obtained by different observation methods. By constructing a disturbance feature matrix, a three-axis nine-element classification system of "phenomenon axis-structure axis-mechanism axis" is established to improve the accuracy of disturbance / inhomogeneity classification. A deep learning model is introduced for intelligent identification and classification training of disturbance types.

[0004] The present invention can solve the problem that the existing ionospheric detection method mainly relies on a single data source for classification, thereby improving the detection accuracy and classification capability of ionospheric disturbances.

[0005] To achieve the above-mentioned object, the present invention provides a multi-method collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities, comprising the following steps:

[0006] Step 1: Deploy a multi-source ionospheric observation system in the low-latitude region to identify low-latitude ionospheric disturbances / inhomogeneities and obtain multi-source observation data;

[0007] Step 2: Based on the multi-source observation data obtained in step 1, a multi-physics field coupling model is constructed to analyze the physical mechanism of ionospheric disturbances caused by different disturbance sources and analyze low-latitude ionospheric disturbances / inhomogeneities.

[0008] Step 3: Integrate the multi-source observation data from step 1 and the multi-physics field coupling model from step 2 to construct a classification system based on the "phenomenon axis-structure axis-mechanism axis" standard, and construct a classification model based on the disturbance feature matrix, combined with deep learning to classify low-latitude ionospheric disturbances / inhomogeneities.

[0009] Optionally, the multi-source ionospheric observation system consists of a GNSS TEC / scintillation monitor, an ionospheric altimeter and a multi-radar system, wherein the GNSS TEC / scintillation monitor is used to obtain the total electron content, scintillation index and phase scintillation information of the ionosphere, and monitor the intensity and spatial distribution of ionospheric disturbances; the ionospheric altimeter is used to measure the critical frequency and peak height of the ionospheric F2 layer; the multi-radar system includes a low-latitude over-the-horizon ionospheric radar, a VHF coherent scattering radar and an all-sky radar, wherein the low-latitude over-the-horizon ionospheric radar is used to monitor ionospheric inhomogeneities >1000km, the VHF coherent scattering radar is used to detect 10-1000km inhomogeneities, and the all-sky radar is used to monitor ionospheric disturbances <10km.

[0010] Optionally, a modified model based on power spectrum density is proposed for the scintillation index S4 in step 1 to distinguish turbulence-dominated inhomogeneities from layered inhomogeneities. The expression is as follows:

[0011]

[0012] where P(k) is the power spectral density of the TEC amplitude scintillation; k is the wave number; and the exponent β represents the spectral morphology, which is used to distinguish different types of inhomogeneities.

[0013] Optionally, the multi-physics field coupling model comprehensively considers solar activity, geomagnetic activity, neutral atmospheric dynamics and electromagnetic field effects, and uses multi-source observation data parameters to analyze the impact of solar activity and geomagnetic activity on ionospheric disturbances and scintillation in low-latitude regions;

[0014] The expression of the improved TEC change rate model in the multiphysics coupling model is as follows:

[0015]

[0016] Among them, τ s is the solar activity response time constant, τ d is the time constant of geomagnetic activity response, which is used to correct the different response lags of TEC to solar activity and geomagnetic activity; is the TEC change rate in the latitudinal direction, is the reference latitude gradient.

[0017] Optionally, the Hilbert-Huang transform is introduced into the multi-physics coupling model to perform spectral analysis on the VTEC data of a solar cycle to capture the nonlinear periodic characteristics of TEC changes. The expression is as follows:

[0018]

[0019] Among them, the IMF i (f) is the i-th intrinsic mode function, which is used to decompose the different time scale perturbations of TEC changes; R(f) is the residual term, which characterizes the non-periodic changes.

[0020] Optionally, in step 3, a mathematical model of a three-axis nine-element classification standard is constructed, where the phenomenon axis includes flicker scale, morphology, and time-varying characteristics. The model is as follows:

[0021] Φ={S4 spec ,ΔTEC * ,λ h}

[0022] Among them, S4 spec is the S4 correction model in step 1, ΔTEC * is the TEC rate of change model in step 2, λ h is the wavelength of the gravity wave in step 2; the structural axis includes configuration-spectral characteristics-motion characteristics, and the model is as follows:

[0023] Λ={L scale ,δ v ,v E×B}

[0024] Among them, L scale represents the typical spatial scale of the perturbation structure, δ vrepresents the spectrum width of the radar, V E×B Indicates E×B drift speed;

[0025] The mechanism axis includes driving source-energy path-feedback mechanism, and the model is as follows:

[0026]

[0027] Among them, E p is the electric field penetration strength in step 2; is the Pedersen conductivity in step 2; is the electron pressure gradient.

[0028] Optionally, a classification model based on a perturbation feature matrix is ​​also proposed in step 3. The expression of the perturbation feature matrix X is as follows:

[0029]

[0030] Among them, the first three items of the feature matrix X {S4 spec ,ΔTEC * ,λ h} from the phenomenon axis Φ; the middle three items {L scale ,δ v ,v E×B} comes from the structural axis Λ; the last three terms From the mechanism axis ψ.

[0031] The present invention provides a multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities. By constructing a multi-source detection system and adopting a "multi-network collaboration-multi-source fusion" observation mode, full spatiotemporal coverage of low-latitude ionospheric disturbances is achieved. In addition, a data fusion method based on physical constraints is used to construct an "ionosphere-middle atmosphere-troposphere" multi-physical field coupling mechanism to optimize the spatiotemporal consistency of data obtained by different observation methods. Finally, a multi-scale signal processing method is used to extract key feature parameters, and a "three-axis nine-element" disturbance / inhomogeneity classification standard is proposed in combination with radar observation data (such as drift velocity and spectral width). A disturbance feature matrix is ​​constructed, and a classifier is trained in combination with a deep learning model (CNN, LSTM) to achieve accurate classification of low-latitude ionospheric disturbances and improve the GNSS signal anomaly recognition capability. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0033] Figure 1 The present invention is a schematic block diagram showing the principle of a multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities.

[0034] Figure 2 It is a schematic diagram of the execution flow of step 1 in the multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities of the present invention.

[0035] Figure 3 It is a schematic diagram of the execution flow of step 2 in the multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities of the present invention.

[0036] Figure 4 It is a schematic diagram of the execution flow of step 3 in the multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities of the present invention. DETAILED DESCRIPTION

[0037] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0038] The present invention provides a multi-method collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities, comprising the following steps:

[0039] Step 1: Deploy a multi-source ionospheric observation system in the low-latitude region to identify low-latitude ionospheric disturbances / inhomogeneities and obtain multi-source observation data;

[0040] Step 2: Based on the multi-source observation data obtained in step 1, a multi-physics field coupling model is constructed to analyze the physical mechanism of ionospheric disturbances caused by different disturbance sources and analyze low-latitude ionospheric disturbances / inhomogeneities.

[0041] Step 3: Integrate the multi-source observation data from step 1 and the multi-physics field coupling model from step 2 to construct a classification system based on the "phenomenon axis-structure axis-mechanism axis" standard, and construct a classification model based on the disturbance feature matrix, combined with deep learning to classify low-latitude ionospheric disturbances / inhomogeneities.

[0042] The technical structure principle block diagram of the method of the present invention is as follows Figure 1 As shown, the following is further explained in combination with the specific implementation process:

[0043] The specific implementation of step 1 is as follows:

[0044] A multi-source ionospheric observation system will be deployed in low-latitude regions, including "GNSS TEC / scintillation monitors, ionospheric altimeters, and multi-radar systems." The "multi-radar system" includes but is not limited to "low-latitude over-the-horizon ionospheric radar, VHF coherent scatter radar, and all-sky radar."

[0045] ①GNSS TEC / scintillation monitor: used to obtain TEC values, scintillation index (S4), and phase scintillation information.

[0046] ②Multi-radar system:

[0047] Low-latitude over-the-horizon ionospheric radar: used to monitor large-scale (>1000km) ionospheric inhomogeneities, such as traveling ionospheric disturbances (TIDs). VHF coherent scatter radar: used to detect medium-scale (10-1000km) inhomogeneities, such as equatorial plasma bubbles (EPBs). All-sky radar: used to monitor small-scale (<10km) ionospheric disturbances, such as Kelvin-Helmholtz instability (KHIs).

[0048] ③ Ionospheric altimeter: used to measure the foF2 (critical frequency) and hmF2 (peak height) of the ionosphere and provide ionospheric profile data.

[0049] The execution process of the multi-source ionospheric observation system in step 1 can be found in Figure 2 , which is described below with reference to specific embodiments:

[0050] First, a GNSS TEC / scintillation monitor is used to obtain TEC values, scintillation index (S4), and phase scintillation information.

[0051] Three GNSS TEC / scintillation monitors were deployed in the low-latitude region, observing along three directions: 25°N, 45°N, and 115°E. The TEC value was calculated using the observation data from the GNSS TEC / scintillation monitors. The traditional TEC calculation method uses the carrier phase smoothed pseudorange method, and its expression is:

[0052]

[0053] The traditional TEC calculation formula does not fully consider the influence of the background electric field (such as the easterly electric field EEE) and the equatorial fountain effect. Therefore, the present invention proposes to introduce a background electric field correction term and an equatorial electric field-drift term to optimize the TEC calculation model:

[0054]

[0055] Where E×B represents the electric field drift term, reflecting the longitudinal drift effect of the low-latitude ionosphere; v di is the ion drift velocity, v dnis the neutral wind driving speed, and the coupling of the two determines the TEC change rate; h m F2 is the peak height of the F layer, and h0 is the height of the measuring station.

[0056] After obtaining the new TEC calculation model, a multi-scale wavelet transform (WT) is introduced to identify disturbance components of different scales and construct a multi-scale ROTI index:

[0057]

[0058] Among them, s is the disturbance scale, ws is the wavelet weight factor, which determines the contribution of disturbances of different scales; ROT s represents the ROT calculated values ​​at different scales; this formula distinguishes the effects of large-scale traveling disturbances (TIDs) and small-scale plasma bubbles (EPBs) on TEC changes.

[0059] The traditional S4 index only reflects the signal amplitude scintillation intensity, without considering the background electric field effect and the difference in disturbance type. It is difficult to distinguish scintillation phenomena caused by different mechanisms such as plasma bubbles (EPBs) and small and medium-scale traveling disturbances (TIDs). In order to improve the physical representativeness and classification and discrimination ability of the S4 index, it is necessary to introduce spectral structure information to enhance its sensitivity to disturbance types. Therefore, in step 1, the present invention also proposes a S4 correction model based on power spectral density (PSD) to distinguish turbulence-dominated inhomogeneities (Kolmogorov cascades) from layered structure inhomogeneities (gradient drift instability):

[0060]

[0061] where P(k) is the power spectral density of the TEC amplitude scintillation; k is the wave number; and the exponent β represents the spectral morphology and is used to distinguish different types of inhomogeneities: if β≈5 / 3, the scintillation is mainly dominated by turbulent processes (Kolmogorov cascades); if β≈2 or higher, it may be caused by gradient drift instability.

[0062] Secondly, through a multi-radar system, including but not limited to "low-latitude over-the-horizon ionospheric radar, VHF coherent scattering radar, all-sky radar", etc., based on the interferometric positioning measurement method, by measuring the actual phase difference of the echo of the inhomogeneous body:

[0063] φ ij =Φ ij +360·m+φ biasij (5)

[0064] Φ ij (-180° to 180°) is the phase difference calculated by the cross-correlation technique, φbiasij is the system phase offset, ij represents different radar antenna arrays, and the elevation cosine is calculated jointly based on the baseline lengths between arrays of different lengths on the same horizontal line:

[0065]

[0066] The unambiguous target elevation angle θ1 on this baseline can be obtained. Similarly, by calculating the elevation angle θ2 of the target on another perpendicular baseline, the true arrival angle information of the target can be located according to the spatial geometric relationship:

[0067]

[0068] α and θ represent the azimuth and elevation angles of the target obtained by the final positioning, respectively. The target distance is calculated based on the radar echo time difference:

[0069] R=c·Δt / 2 (10)

[0070] By solving the large triangle in space with the radius of the earth as the side length, the height component h and horizontal distance of the target can be obtained:

[0071]

[0072] d z =Rcosθsinα (12)

[0073] d m =Rcosθcosα (13)

[0074] In addition to applying the interferometric positioning principle, the developed multi-base all-sky radar adopts multi-base positioning technology by solving the three-dimensional space vector:

[0075] R1=R2-r (14)

[0076] R1, R2, r represent the target distance vector and the distance vector between the two bases respectively. After de-noising the original radar echo data, the echo signal-to-noise ratio SNR, Doppler velocity v and spectrum width δv are calculated, and the E×B drift velocity V can also be obtained. E×B (plasma drift velocity in the direction orthogonal to the electric field and magnetic field), which can comprehensively monitor the spatial structure changes and movement processes of inhomogeneous bodies.

[0077] Low-latitude over-the-horizon ionospheric radar: used to monitor large-scale (>1000km) ionospheric inhomogeneities, such as traveling ionospheric disturbances (TIDs). VHF coherent scatter radar: used to detect medium-scale (10-1000km) inhomogeneities, such as equatorial plasma bubbles (EPBs). All-sky radar: used to monitor small-scale (<10km) ionospheric disturbances, such as Kelvin-Helmholtz instability (KHIs).

[0078] For large-scale ionospheric spatial disturbances, a joint inversion system for large-scale ionospheric disturbances / inhomogeneities is constructed based on the hemispheric coverage detection capability of over-the-horizon radar and combined with GNSSTEC / scintillation monitors. Over-the-horizon radar (OTHR) uses multiple bounce reflections of electromagnetic waves in the ionosphere and the ground to achieve spatial position calibration of large-scale ionospheric disturbances (such as traveling disturbance TIDs and magnetospheric energy injection events). The phase change of its echo signal can be used to invert the disturbance characteristics of the ionospheric plasma density, and the "echo phase disturbance" is used to calculate the ionospheric disturbance intensity:

[0079]

[0080] Where Δφ is the phase change of the radar echo; λ is the wavelength of the radar signal; N e (h) is the electron density distribution; h0 and h F , are the lowest and highest heights of the reflection points respectively.

[0081] For intermediate-scale disturbances / inhomogeneities, interferometric positioning is performed using VHF coherent scatter radar to locate their spatial position, achieving three-dimensional reconstruction of the ionospheric inhomogeneity. The interferometric positioning results are then used to accurately define the penetration point of the TEC / scintillation position. The interferometric positioning results are then fused and corrected with the geometric projection of the GNSS penetration point, and a dynamic position correction algorithm for the penetration point is established. This suppresses TEC measurement errors caused by the horizontal gradient of the ionosphere and improves the signal drift calibration accuracy of the scintillation receiver. With the help of VHF coherent scatter radar, high-precision positioning can be achieved through interferometric measurement technology. The typical growth rate of EPB is given by the generalized Rayleigh-Taylor model:

[0082]

[0083] Where g is the acceleration due to gravity; ν in is the ion-neutral particle collision frequency; k ⊥ is the wave vector perpendicular to the magnetic field; ρ i is the ion cyclotron radius. The three-dimensional spatial distribution of the inhomogeneous body can be obtained by radar interferometry, and its height component h is calculated by the following formula:

[0084]

[0085] Where R is the target distance measured by the radar; R E is the radius of the Earth; θ is the elevation angle measured by the radar. Furthermore, the EPB structure is further calibrated using GNSS TEC / scintillation observations. The interferometric positioning results are fused with the GNSS penetration point geometric projection to establish a dynamic position correction algorithm for the penetration point, suppressing TEC measurement errors caused by ionospheric horizontal gradients.

[0086] For small-scale disturbances / inhomogeneities, all-sky radars are used to analyze the fine structure of the ionosphere through Doppler beam sharpening technology and space-time adaptive processing algorithms. Based on interferometric positioning and multi-base monitoring technology, the drift characteristics of disturbances / inhomogeneities in the local area can be further analyzed to correct the small-scale positioning results of GNSS scintillation. At the same time, the radar system's excellent vertical observation capability from the lower atmosphere to the ionosphere provides observational data for the study of the correlation between the low-level atmospheric disturbance sources of small-scale ionospheric disturbances and the fading mechanism of GNSS scintillation signals. Small-scale inhomogeneities (such as ionospheric turbulence structures) are mainly triggered by Kelvin-Helmholt instability (KHI). Using all-sky radars, the fine structure can be analyzed through Doppler beam sharpening technology and space-time adaptive processing algorithms. The KHI triggering conditions are:

[0087]

[0088] Where Ri is the Richardson number; g is the acceleration due to gravity; and dU / dz is the vertical wind shear. When Ri < 0.25, KHI occurs, forming turbulent inhomogeneities, which affect the phase and amplitude scintillation of the GNSS signal. The turbulent characteristics of TEC oscillations are analyzed using power spectral density (PSD). The Kolmogorov cascade exponent β is calculated to distinguish turbulence-dominated inhomogeneities from layered inhomogeneities. For details, refer to Equation (4) in Step 1.

[0089] Finally, the ionospheric altimeter was used to scan and obtain the ionospheric frequency-height map, calibrate and extract the foF2 and hmF2 parameters, and obtain the foF2 mutation events.

[0090] The ionospheric altimeter uses vertical incidence pulse wave measurement technology. By emitting radio waves of different frequencies and recording their return time, it constructs the wave propagation path and ultimately forms a frequency-height map of the ionosphere. The altimeter's workflow is as follows: Radio waves from 1 MHz to 30 MHz are emitted in sequence, and the time delay, amplitude, and polarization characteristics of the echo signal are recorded. The virtual height is calculated based on the propagation time of the echo, and a preliminary ionospheric profile is formed. The electron density profile inversion algorithm (such as ARTIST or SAO Explorer software) is used to convert the virtual height into the real height and optimize the ionospheric structure model. Based on the altimeter scanning data, an ionospheric frequency-height map is generated for subsequent parameter extraction and event detection.

[0091] In the process of foF2 parameter extraction, calibration is performed based on the strongest echo frequency of the F2 layer in the frequency-height map, and the least squares fitting algorithm is used to remove the influence of the E layer and F1 layer to obtain a stable and reliable foF2 value. The calculation formula of foF2 is as follows:

[0092] foF2=f max (E,F1,F2) (19)

[0093] At the same time, in order to further verify the accuracy of foF2, it can be cross-validated with GNSS TEC data and compared with the ionospheric background model. In addition, foF2 can be used to calculate the peak electron density NmF2 of the F2 layer, and its calculation formula is as follows:

[0094] NmF2=1.24×10 10 ×(foF2) 2 (20)

[0095] Among them, the unit of foF2 is MHz, and the unit of NmF2 is electrons / cm 3 , this parameter can be used to characterize the overall variation trend of the ionospheric electron density.

[0096] In the process of extracting hmF2, the true height inversion algorithm is used to calculate the height corresponding to the peak electron density of the F2 layer. The calculation accuracy of the peak height is optimized by combining the altimeter echo delay information and the electron density gradient change. The calculation formula of hmF2 is as follows:

[0097]

[0098] in, is the echo delay corresponding to the maximum frequency, and the true height value is obtained through the ionospheric profile inversion algorithm. To further improve the measurement accuracy of hmF2, it can be corrected using GNSS TEC data and calibrated in combination with empirical ionospheric models (such as the IRI model).

[0099] In the process of detecting foF2 mutation events, we first conduct time series analysis on the altimeter observation data. The short-time Fourier transform (STFT) method is used to calculate the short-time spectrum changes of foF2. The wavelet transform (WT) is then combined to identify disturbances on different time scales to distinguish short-term disturbances (such as traveling ionospheric disturbances (TIDs)) from long-term background changes. In addition, to accurately determine the mutation events of foF2, the change rate of foF2 is calculated based on the threshold judgment method. The calculation formula is as follows:

[0100]

[0101] When ΔfoF2 exceeds a set threshold (e.g., 0.5 MHz / h), it is identified as a mutation event and a secondary confirmation is performed based on the changing trend of hmF2 to ensure the reliability of the test results. Furthermore, machine learning methods, such as the LSTM neural network model, are used to train historical datasets to predict temporal changes in foF2 and identify outliers, thereby improving the accuracy of mutation event detection.

[0102] The specific implementation of step 2 is as follows:

[0103] The multi-source observation data obtained in step 1 must first be preprocessed and fused in order to construct the multi-physics coupling model in step 2:

[0104] For the transmission and preprocessing of GNSS TEC / scintillation monitor data, the corrected TEC value (TEC corr ) is used as the basic input of the background ionospheric electron density in step 2 to analyze the impact of solar activity and geomagnetic activity on low-latitude ionospheric disturbances; the multi-scale ROTI index (ROTI) calculated by formula (3) in step 1 is used to calculate the ROTI index. MS ) is directly input into step 2 to quantify the intensity and spatial distribution characteristics of disturbances of different scales; the modified scintillation index S4 calculated by formula (4) in step 1 is converted into spec It is passed to step 2 to distinguish turbulence-dominated inhomogeneities from layered inhomogeneities, and serves as an important input for constructing the disturbance characteristic matrix.

[0105] Regarding the transmission and processing of multi-radar system data, the echo phase disturbance Δφ data obtained by the low-latitude over-the-horizon ionospheric radar in step 1 is transmitted to step 2 for improving the calculation of the TEC change rate model; the inhomogeneous body interferometric positioning data (position, height, and morphology) measured by the VHF coherent scattering radar is transmitted to step 2 for analyzing the three-dimensional spatial distribution and dynamic evolution process of the EPB.

[0106] Regarding the transmission and utilization of ionospheric altimeter data, the foF2 (critical frequency) and hmF2 (peak height) data measured in step 1 are transmitted to step 2 to constrain the vertical structure of the ionosphere; the ionospheric altimeter frequency-height map data is used in step 2 to verify the rationality of the vertical distribution of the multi-physics field coupling model.

[0107] The "ionosphere-middle atmosphere-troposphere" multi-physics coupling model comprehensively considers solar activity, geomagnetic activity, neutral atmospheric dynamics and electromagnetic field effects to analyze the driving effect of different disturbance sources on ionospheric disturbances. The multi-source observation data obtained in step 1 are used to establish the "ionosphere-middle atmosphere-troposphere" multi-physics coupling model. The key parameters such as the monthly mean TEC (TECmon), F10.7 correction index (F10.7adj), sunspot number (SSN), Dst index and Kp index obtained by the multi-source ionospheric observation system in step 1 are used to analyze the impact of solar activity and geomagnetic activity on ionospheric disturbances and scintillation in low-latitude regions. Among them, observation data with an altitude greater than 80km are used for the spatiotemporal evolution of the ionosphere, observation data with an altitude between 20km and 80km are used for the spatiotemporal evolution of the middle atmosphere, and observation data with an altitude less than 20km are used for the spatiotemporal evolution of the troposphere. The specific process is as follows. Figure 3 shown.

[0108] The impact of solar and geomagnetic activity on ionospheric disturbances and scintillation in low-latitude regions is analyzed by combining the monthly mean TEC value (TECmon), the F10.7 correction index (F10.7adj), SSN, Dst, and the Kp index. To deepen the study of TEC variations in the low-latitude ionosphere, a physically constrained normalized TEC variation rate model and a nonlinear coupling equation driven by solar-geomagnetic activity are introduced on the basis of the original statistical analysis to more accurately characterize the evolution of TEC under different solar activity levels. Based on the latitude gradient effect, the delayed response of solar activity changes, and the dynamic influence of geomagnetic activity, an improved TEC variation rate model is proposed:

[0109]

[0110] Among them, τ s is the solar activity response time constant, τ d is the time constant of geomagnetic activity response, which is used to correct the different response lags of TEC to solar activity and geomagnetic activity; is the TEC change rate in the latitudinal direction, is the reference latitudinal gradient.

[0111] Furthermore, the long-term VTEC data obtained in step 1 are processed, the diurnal VTEC variations are classified, and the spectrum of the VTEC data over a solar cycle is analyzed using a fast Fourier transform to capture the precise periodic information implicit in the interannual VTEC variations. The Hilbert-Huang transform (HHT) is introduced into the VTEC spectrum analysis to better capture the nonlinear periodic characteristics of TEC variations. The model is as follows:

[0112]

[0113] Among them, the IMF i (f) is the i-th intrinsic mode function (IMF), which is used to decompose the different time scale perturbations of TEC changes; R(f) is the residual term, which characterizes the non-periodic changes.

[0114] By analyzing the curve variation patterns and correlation coefficient ρ between the monthly mean VTEC and the solar activity index SSN, F10.7, geomagnetic activity index Dst, Kp, the impact of solar and geomagnetic activities on the background ionosphere in low-latitude areas is analyzed.

[0115] The interferometric positioning results of the inhomogeneous body obtained by the multi-radar system in step 1, including parameters such as the target elevation angle θ, azimuth angle α, and height h, are used to analyze the field / non-field characteristics, continuous / quasi-periodic structural characteristics, and the performance characteristics of the inhomogeneous body at different heights (thin layer at the bottom of the F zone, extended F, plasma bubbles, plasma blocks, etc.), as well as the structural evolution and drift motion over time.

[0116] The specific physical mechanism coupling process is as follows:

[0117] The three-dimensional wave vector components can be obtained by bistatic radar interferometry. The non-field-oriented structure is manifested as enhanced scattering away from the direction of magnetic field lines. Its appearance is related to the gradient drift instability, and its growth rate can be expressed as:

[0118] γ=(g / νin)(k⊥2ρi2) / (1+k⊥2ρi2) (25)

[0119] Where g is the effective gravitational acceleration, νin is the ion-neutral particle collision frequency, and ρi is the ion cyclotron radius. The motion velocity is calculated using the time-shifted cross-correlation algorithm:

[0120]

[0121] The EF coupling process can be analyzed by solving the current continuity equations of the E and F layers. Similar to the EPB process in low-latitude regions, it is necessary to couple the growth rate of the generalized Rayleigh-Taylor instability:

[0122]

[0123] Among them H p =k B T / (m P g) is the pressure elevation, v in is the ion-neutral particle collision probability, and considering the electric field penetration depth dominated by the Pedersen current, the electric field / current effect in the EF coupling process is analyzed:

[0124]

[0125] in is the Pedersen conductivity. By superimposing the wind field, the neutral wind shear driving process can also be coupled, which also takes into account the modulation effect of gravity wave propagation on the inhomogeneous body:

[0126]

[0127] β is the composite coefficient, ω is the gravity wave frequency, v n is the neutral wind speed, and the linearized gravity wave governing equation is:

[0128]

[0129] Where ω' is the vertical disturbance velocity, p' is the pressure disturbance, The gravity wave energy input process can be calculated based on the TID horizontal phase velocity v h The wavelength is calculated with the period T to obtain:

[0130]

[0131] Where H is the atmospheric scale height, g′=g(ΔT / T)g′=g(ΔT / T) is the reduced gravity. The neutral wind field is obtained by inverting the Doppler velocity of the meteor wake:

[0132] v D =2(v m ·k r ) / |k r | (34)

[0133] where v m is the meteoroid velocity, k r is the radar wave vector. When the vertical wind shear satisfies the KH instability growth rate condition, the growth rate can be expressed as follows:

[0134] ν KH =(ΔU / Δz) 0.5 (kΔU 2 / g') 0.5 (35)

[0135] Where ΔU is the velocity shear, g' is the reduced gravitational acceleration, and k is the horizontal wave number. According to the shear layer vorticity evolution equation:

[0136]

[0137] Analyze the key role of wind shear mechanism and KH instability, among which is the vorticity, v is the kinematic viscosity, and the coupling strength parameter is defined to quantify the electromagnetic field-neutral wind interaction efficiency:

[0138]

[0139] In the extended F region, the density gradient couples with the electric field to produce instability, and its critical condition is:

[0140]

[0141] Where μ is the plasma mobility and E0 is the background electric field. The process can be described by the continuity equation:

[0142]

[0143] Where V is the plasma drift velocity, and S and L are the generation / loss terms. Considering the combined influence of the penetrating electric field and the generator electric field in the low-latitude F region, the competitive relationship between the two needs to be coupled. The empirical formula for the penetrating electric field strength can be expressed as:

[0144]

[0145] Where Φ PC is the electric potential, Σ P is the Pedersen conductivity integral, and τ is the decay time. The generator electric field is represented by the neutral wind driving source:

[0146]

[0147] where v n is the neutral wind speed, P e is the electron pressure. Taking into account the neutral component disturbance and the ionospheric storm response, such as the increase in the F region recombination rate caused by the decrease in the O / N2 ratio during the negative storm:

[0148] β eff =k1[O2]+k2[N2] (42)

[0149] In order to distinguish the effects of various physical quantities on different disturbance / inhomogeneous body phenomena, principal component analysis (PCA) is used to analyze multiple parameters (such as ΔNe, S4, v n , E×B drift, etc.) to construct the covariance matrix:

[0150]

[0151] At the same time, the Bayesian inversion framework is applied, and the likelihood function is:

[0152]

[0153] θ includes parameters such as the electric field, wind field, and conductivity. The posterior distribution is solved through Markov chain Monte Carlo sampling, and then the contribution rate of each mechanism is quantified through eigenvalue decomposition, constructing a weight allocation method for low-latitude ionospheric disturbance / inhomogeneity mechanism.

[0154] By coupling the above-mentioned physical processes such as electric fields, wind fields, gravity waves, and background electron density changes, combined with TEC disturbances and scintillation observed by GNSS, and altimeter frequency-height maps, multi-source data are jointly constrained to analyze the various physical processes behind various ionospheric anomalies, such as ionospheric traveling disturbances, positive and negative storm responses, abnormally enhanced or depleted structures, EPBs, and super EPBs when the sun enters an active period. A multi-physics field coupling model of "ionosphere-middle atmosphere-troposphere" is constructed to analyze the driving effect of different disturbance sources on ionospheric disturbances, providing a basis for further analysis of the impact of these phenomena on the ionosphere.

[0155] The specific implementation of step 3 is as follows:

[0156] Construct a "three-axis nine-element" classification system based on the "phenomenon axis-structure axis-mechanism axis", and combine it with machine learning technology for automatic classification. The multi-observation data obtained in step 1 are systematically integrated to provide a comprehensive feature description of the disturbance, providing a data basis for the classification system in step 3; the contribution weights of different physical processes such as solar activity, geomagnetic activity, and neutral wind field to the disturbance quantified in step 2 assign differentiated weights to each element of the characteristic matrix in step 3, including the proportion weights of generator electric field, rapid penetration electric field, neutral wind field, neutral component disturbance, gravity wave, EF zone electromagnetic coupling, RT instability, lower atmosphere-ionosphere electromagnetic coupling and energy transport, etc.; the multi-physics field coupling model in step 2 dynamically adjusts the classification results in step 3, and the physical driving factors (such as electric field, neutral wind, etc.) obtained by inversion based on ionospheric imaging data in step 2 can be used as key classification features in step 3. The execution process is as follows: Figure 4 shown.

[0157] The formation process of disturbances / inhomogeneities in the low-latitude ionosphere involves electromagnetic coupling of the EF layer, Rayleigh-Taylor (RT) instability, and gravity wave-plasma interaction. Traditional classification methods fail to fully describe these driving mechanisms. Therefore, a nonlinear disturbance growth rate equation is introduced:

[0158]

[0159] Among them, Γ RT is the Rayleigh-Taylor instability growth rate correction term; Γ KH is the Kelvin-Helmholtz (KH) instability term, which characterizes the turbulence enhancement induced by wind shear; Γ GW is the gravity wave modulation term, which describes the influence of atmospheric gravity waves on plasma disturbances; Γ E-F is the electromagnetic coupling term of the EF layer, describing the interaction between the electric field and the neutral wind; α r is the plasma dissipation term, which characterizes the suppression effect of the plasma recombination process on the disturbance.

[0160] The equatorial plasma bubble (EPB) in the low-latitude ionosphere is mainly triggered by RT instabilities, but the traditional RT model does not fully consider the background electric field-wind field coupling and turbulence enhancement effects. Therefore, the following improved model is proposed:

[0161]

[0162] in, is the wind-electric field coupling term, which describes the enhancement effect of neutral wind and shear on RT instability; is the vertical electric field gradient suppression term, which characterizes the modulation effect of the penetration depth of the electric field in the EF layer on the RT development process; C t ∈ 1 / 3 k 2 / 3 is the turbulence enhancement term, which describes the modification of the RT growth rate by the turbulence cascade.

[0163] A classification system for ionospheric disturbances at low latitudes is established, integrating phenomenological characteristics, structural parameters and physical mechanisms. According to the scintillation scale, range, duration, periodicity, amplitude, phase and other phenomena, the corresponding typical / infrequent ionospheric disturbances and inhomogeneous body types, positions, scales, structures, motions, periods and other characteristics are correlated, and the inherent physical mechanisms of the associated phenomena are further explored, such as wind shear, polarized electric field, equatorial fountain effect, SSN, Dst, EF coupling and troposphere-ionosphere coupling.

[0164] By classifying disturbances and inhomogeneities that affect ionospheric anomalies, a refined classification database is constructed. Based on the above classification system, a "three-axis nine-element" classification standard is proposed: the phenomenon axis: scintillation scale-morphology-time-varying characteristics; the structural axis: configuration-spectral characteristics-motion characteristics; and the mechanism axis: driving source-energy path-feedback mechanism.

[0165] Constructing a mathematical model for classifying low-latitude ionospheric disturbances / inhomogeneities, this paper proposes a classification model based on the disturbance characteristic matrix in order to more accurately describe different types of ionospheric anomalies and their impact on GNSS signals. Define the disturbance characteristic matrix X:

[0166]

[0167] Among them, the first three items of the feature matrix X {S4 spec ,ΔTEC * ,λ h} from the phenomenon axis Φ; the middle three items {L scale ,δ v ,v E×B} comes from the structural axis Λ; the last three terms Derived from the mechanism axis ψ; each item is a physically observable quantity or a dynamically invertible quantity with a clear physical meaning; it can be used as the input feature vector of the machine learning model, the basis for generating disturbance classification labels, and can also be used for disturbance prediction and causal modeling.

[0168] (1) The phenomenon axis Φ is “scintillation scale-morphology-time-varying characteristics”, and the model is as follows:

[0169] Φ={S4 spec ,ΔTEC * ,λ h} (48)

[0170] S4 spec is the S4 correction model in step 1, and its expression refers to formula (4);

[0171] ΔTEC * is the TEC change rate model in step 1, and its expression refers to formula (23);

[0172] λ h is the wavelength of the gravity wave in step 2, expressed as:

[0173]

[0174]

[0175] (2) The structural axis Λ “position-spectral characteristics-motion characteristics”, the model is as follows:

[0176] Λ={L scale ,δ v ,v E×B} (50)

[0177] L scale The typical spatial scale of the disturbance structure is expressed as:

[0178] L scale =Vd *Γ rt (51)

[0179] Among them, V d represents the disturbance drift velocity, which can be obtained by multi-radar system speed measurement; Γ rt represents the growth time scale of the Rayleigh-Taylor instability, which can be obtained from formula (46) in step 3:

[0180] δ v represents the spectral width of the radar, v E×B Indicates E×B drift speed, both parameters can be obtained through the multi-radar system.

[0181]

[0182] (3) Mechanism axis ψ “driving source-energy path-feedback mechanism”, the model is as follows:

[0183]

[0184] E p is the penetration electric field strength in step 2, which can be obtained by formula (40);

[0185] is the Pedersen conductivity in step 2, which can be obtained by formula (30);

[0186] is the electron pressure gradient, which is expressed as:

[0187]

[0188] Among them, k B is the Boltzmann constant, 1.38×10 -23 J / K;T e is the electron temperature, obtained by the IRI model; The spatial gradient of electron density can be obtained by collecting TEC data using a GNSS TEC / scintillation monitor and constructing a three-dimensional structure of the electron density.

[0189]

[0190] Through the "phenomenon axis-structure axis-mechanism axis" classification standard, each type of disturbance sample is mapped to a corresponding nine-element label as the target output of supervised learning. Using labeled samples, classification models (such as random forest, XGBoost, support vector machine, etc.) are trained to realize intelligent identification of disturbance types. In order to further improve the modeling ability of the disturbance evolution process, time series deep learning networks such as LSTM are introduced to model the changes of disturbance characteristics over time, and identify disturbance development trends and potential anomalies. In addition, unsupervised methods such as K-means clustering can be applied to unlabeled historical data, combined with the feature matrix X to perform clustering identification of disturbance mechanisms, and assist the expert system in semi-supervised learning training. The entire process realizes a complete closed loop from GNSS observation data, disturbance feature extraction, three-axis mapping, model training to disturbance classification, which improves the accuracy, automation and physical interpretation capabilities of disturbance identification.

[0191] In summary, the multi-method collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities proposed in the present invention has the following beneficial effects:

[0192] First, by integrating multiple detection methods, including over-the-horizon radar, ionospheric altimeter, GNSS TEC / scintillation monitor, coherent scatter radar, and multi-base all-sky radar, the present invention achieves cross-scale, full-time and full-space ionospheric disturbance monitoring. This overcomes the limitations of a single observation method and significantly improves the spatial coverage and temporal continuity of observation data. For example, GNSS TEC monitoring can provide a wide range of ionospheric total electron content (TEC) variation information, coherent scatter radar can detect the spatial structure of plasma bubbles (EPBs) and turbulent inhomogeneities with high resolution, and over-the-horizon radar can be used to detect large-scale ionospheric disturbances (such as traveling disturbances (TIDs)). Together, these methods form a complementary observation network, improving the accuracy and reliability of detecting low-latitude ionospheric disturbances.

[0193] Secondly, the present invention deeply analyzes the driving mechanism of ionospheric disturbances by constructing a multi-physics field coupling model of "ionosphere-middle atmosphere-troposphere", and quantifies the contribution of factors such as solar activity, magnetic storms, wind shear, and gravity waves to ionospheric disturbances. Compared with traditional empirical modeling methods, this scheme adopts Bayesian inversion technology and numerical simulation methods, which can more accurately trace the origin of disturbances and reveal the complete process of ionospheric inhomogeneities from seed disturbance triggering to nonlinear evolution. In addition, the present invention also uses multi-source data fusion to improve the calculation accuracy of key parameters such as TEC disturbances, scintillation index S4 and phase scintillation, providing a scientific basis for the prediction and early warning of ionospheric disturbances.

[0194] Finally, the present invention proposes a disturbance classification system based on physical constraints, and combines it with machine learning technology to improve the classification accuracy and automation level of ionospheric disturbances. Traditional scintillation classification methods rely on a single parameter and are difficult to distinguish different types of ionospheric anomalies. The present invention constructs a disturbance feature matrix by integrating multidimensional data such as TEC, scintillation index, drift velocity, and spectral width, and uses deep learning models (CNN, LSTM) for classification, thereby improving the ability to identify different types of disturbances. In addition, the present invention also optimizes the GNSS signal anomaly identification method, which can effectively quantify the impact of ionospheric disturbances on navigation and positioning accuracy, thereby providing important support for high-precision navigation, space environment monitoring, and aviation / maritime communications.

[0195] The above disclosure is merely one or more preferred embodiments of the present invention, and certainly cannot be used to limit the scope of the present invention. A person skilled in the art can understand that all or part of the processes of the above embodiments and equivalent changes made in accordance with the claims of the present invention still fall within the scope of the invention.

Claims

1. A multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities, characterized by: The following steps are involved: Step 1: Deploy a multi-source ionospheric observation system in the low-latitude region to identify low-latitude ionospheric disturbances / inhomogeneities and obtain multi-source observation data; Step 2: Based on the multi-source observation data obtained in step 1, a multi-physics field coupling model is constructed to analyze the physical mechanism of ionospheric disturbances caused by different disturbance sources and analyze low-latitude ionospheric disturbances / inhomogeneities. Step 3: Integrate the multi-source observation data from step 1 and the multi-physics field coupling model from step 2 to construct a classification system based on the "phenomenon axis-structure axis-mechanism axis" standard, and build a classification model based on the disturbance feature matrix, combined with deep learning to classify low-latitude ionospheric disturbances / inhomogeneities.

2. The multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities according to claim 1, characterized in that: The multi-source ionospheric observation system includes a GNSS TEC / scintillation monitor, an ionospheric altimeter and a multi-radar system. The GNSS TEC / scintillation monitor is used to obtain the total electron content, scintillation index and phase scintillation information of the ionosphere, and monitor the intensity and spatial distribution of ionospheric disturbances; the ionospheric altimeter is used to measure the critical frequency and peak height of the ionospheric F2 layer; the multi-radar system includes a low-latitude over-the-horizon ionospheric radar, a VHF coherent scattering radar and an all-sky radar. The low-latitude over-the-horizon ionospheric radar is used to monitor ionospheric inhomogeneities >1000km, the VHF coherent scattering radar is used to detect inhomogeneities between 10 and 1000km, and the all-sky radar is used to monitor ionospheric disturbances <10km.

3. The multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities according to claim 2, characterized in that: In step 1, a modified model based on power spectrum density is proposed for the scintillation index S4 to distinguish turbulence-dominated inhomogeneities from layered inhomogeneities. The expression is as follows: where P(k) is the power spectral density of the TEC amplitude scintillation; k is the wave number; and the exponent β represents the spectral morphology, which is used to distinguish different types of inhomogeneities.

4. The multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities according to claim 3, characterized in that: The multi-physics field coupling model comprehensively considers solar activity, geomagnetic activity, neutral atmospheric dynamics and electromagnetic field effects, and uses multi-source observation data parameters to analyze the impact of solar activity and geomagnetic activity on ionospheric disturbances and scintillation in low-latitude regions; The expression of the improved TEC change rate model in the multiphysics coupling model is as follows: Among them, τ s is the solar activity response time constant, τ d is the time constant of geomagnetic activity response, which is used to correct the different response lags of TEC to solar activity and geomagnetic activity; is the TEC change rate in the latitudinal direction, is the reference latitudinal gradient.

5. The multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities according to claim 4, characterized in that: The Hilbert-Huang transform is introduced into the multi-physics coupling model to perform spectral analysis on the VTEC data of a solar cycle to capture the nonlinear periodic characteristics of TEC changes. The expression is as follows: Among them, the IMF i (f) is the i-th intrinsic mode function, which is used to decompose the different time scale perturbations of TEC changes; R(f) is the residual term, which characterizes the non-periodic changes.

6. The multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities according to claim 5, characterized in that: In step 3, a mathematical model of the three-axis nine-element classification standard was constructed, where the phenomenon axis includes the flicker scale, morphology, and time-varying characteristics. The model is as follows: Φ={S4 spec ,ΔTEC * ,l h } Among them, S4 spec is the S4 correction model in step 1, ΔTEC * is the TEC rate of change model in step 2, λ h is the wavelength of the gravity wave in step 2; The structural axis includes configuration-spectral characteristics-motion characteristics, and the model is as follows: Λ={L scale ,d v ,v E×B } Among them, L scale represents the typical spatial scale of the perturbation structure, δ v represents the spectral width of the radar, v E×B Indicates E×B drift speed; The mechanism axis includes driving source-energy path-feedback mechanism, and the model is as follows: Among them, E p is the electric field penetration strength in step 2; ∑p is the Pedersen conductivity in step 2; is the electron pressure gradient.

7. The multi-means collaborative detection and classification method for low-latitude ionospheric disturbances and inhomogeneities according to claim 6, characterized in that: In step 3, a classification model based on the perturbation feature matrix is ​​also proposed. The expression of the perturbation feature matrix X is as follows: Among them, the first three items of the feature matrix X {S4 spec ,ΔTEC * ,λ h } from the phenomenon axis Φ; the middle three items {L scale ,δ v ,v E×B } comes from the structural axis Λ; the last three terms From the mechanism axis ψ.

Images