Virtual transponder capture methods, apparatuses, and media to enhance spoofed jamming identification

Abstract

The application provides a virtual transponder capture method and device based on spoofing interference identification, equipment and medium, wherein the method comprises: collecting positioning observation data of a railway line to construct a positioning observation data set of an interference-free scene; generating a positioning observation augmented data set of a spoofing interference scene based on satellite positioning spoofing interference characteristics and spoofing interference injection tests; training a spoofing interference identification general model in combination with the positioning observation data set and the positioning observation augmented data set, and forming a spoofing interference identification optimization model through an incremental training strategy iterative model training process of augmented data samples; inputting real-time acquired positioning observation data into the spoofing interference identification optimization model to obtain an output result, and judging whether current positioning data in the output result is reliable based on projection residual statistical testing, so as to solve the problem of lack of spoofing interference identification for a virtual transponder capture process.

Description

Technical Field

[0001] The present invention relates to the technical field of railway track transit train operation control, and in particular to a virtual transponder capture method, device, equipment and medium based on deception interference identification. Background Art

[0002] With the development of global satellite navigation systems, virtual transponders based on the positioning function of global navigation satellite systems can replace traditional physical transponders, effectively reducing the installation and maintenance costs of trackside equipment while ensuring good compatibility with existing train operation control systems. However, the inherent vulnerability of satellite navigation systems has largely limited the development and promotion of virtual transponders in their corresponding fields. Since navigation satellites are distributed at an altitude of tens of thousands of miles, the electromagnetic wave signals they transmit are subject to interference from the ionosphere, troposphere, and multipath effects, resulting in a significant reduction in the signal power received by the receiver. Therefore, satellite positioning receivers are extremely vulnerable to signal interference attacks from the surrounding environment in which the receiver is located, which in turn affects the credibility of the capture results of the virtual transponder.

[0003] The types of interference faced by satellite positioning are mainly divided into two categories: deceptive interference and suppression interference. Among them, suppression interference covers satellite navigation signals by releasing higher-power signals, causing receiver performance degradation or even failure, but this interference technology is relatively easy to detect, thereby eliminating the interference effect; deceptive interference uses a format and spectral characteristics similar to real satellite signals to send out fake navigation information with slightly higher power. The receiver will be misled by this deceptive information and then track and lock it. Compared with suppression interference, the implementation method of deceptive interference is more flexible and covert, and poses a greater threat to the satellite navigation system. At the same time, the operation of trains in the vast railway network faces a complex and changeable operating environment, which also brings significant challenges to the satellite navigation system's provision of stable and reliable positioning, navigation and timing services. However, the normal operation of the virtual transponder is highly dependent on the positioning results of the satellite navigation system. Therefore, the identification and detection of satellite navigation deceptive interference during the virtual transponder capture process is an issue that the current railway system needs to consider urgently.

[0004] However, the existing technology generally has the following problems:

[0005] (1) Most existing deception interference detection methods are difficult to apply in the railway field, and there are few studies on deception interference detection methods that directly target the virtual transponder capture process in the train field. Deception interference detection methods assisted by external hardware require the use of multiple receivers or array antennas. However, due to the limited space on the train and the impact of replacing antennas on the overall safety characteristics of the train, this type of method is difficult to implement in reality; deception interference detection technology based on navigation information encryption requires encrypting the navigation information at the signal level to detect the existence of deception interference signals, but this type of method requires changes to the entire signal system, resulting in high costs and difficulty in short-term implementation.

[0006] (2) At present, the research on virtual balises mainly focuses on safety assessment, layout design of virtual balises, specific track database and optimization of capture methods. However, there is little research on protection methods against possible deception interference during the capture of virtual balises and on ensuring the credibility of the capture results of virtual balises. Due to the repetitive nature of train operation, it is possible to introduce machine learning, deep learning and other methods to implement the detection and identification of deception interference during the capture of virtual balises. In addition to satellite navigation positioning receivers, trains also have non-satellite navigation sensors such as axle speed sensors and inertial measurement units that can provide other redundant observation information, which can effectively assist in the detection and identification of deception interference signals during the capture of virtual balises, thereby enhancing the credibility of the capture results of virtual balises.

[0007] Therefore, it is urgent to propose a virtual transponder capture method based on deception interference identification to solve the technical problem that the satellite navigation deception interference detection method lacks deception interference identification for the virtual transponder capture process. Summary of the Invention

[0008] In order to overcome the problems existing in the related art, the present disclosure provides a virtual transponder capture method, device, equipment and medium based on deception interference identification to solve the technical problem of the lack of deception interference identification for the virtual transponder capture process in the related art.

[0009] One or more embodiments of this specification provide a virtual transponder acquisition method based on deception interference identification, which collects positioning observation data of railway lines to construct a positioning observation dataset for interference-free scenarios;

[0010] Generate a positioning observation augmented dataset for deception jamming scenarios based on satellite positioning deception jamming characteristics and deception jamming injection tests;

[0011] Training a general deception interference recognition model by combining the positioning observation data set and the positioning observation amplified data set, and forming a deception interference recognition optimization model by iterating the model training process through an incremental training strategy of the amplified data samples;

[0012] The real-time acquired positioning observation data is input into the deception interference identification optimization model to obtain an output result, and whether the current positioning data in the output result is credible is determined based on a projection residual statistical test.

[0013] Preferably, the collecting of the positioning observation data of the railway line to construct a positioning observation data set for an interference-free scenario specifically includes the following steps:

[0014] Collect railway line measurement data to build line spatial information dataset and virtual balise basic dataset;

[0015] Collect railway line positioning observation data, extract train positioning quality observations after preprocessing, and construct a positioning observation quality assessment set;

[0016] Obtaining neighborhood terrain occlusion boundary feature information of the virtual transponder based on the line spatial information dataset and the virtual transponder basic dataset;

[0017] Determine the neighborhood navigation satellite signal observation performance of the virtual transponder based on the neighborhood terrain occlusion boundary feature information of the virtual transponder and the positioning observation quality assessment set, and construct a neighborhood observation performance feature set of the virtual transponder in an interference-free scenario;

[0018] Based on the train historical observation dataset and the observation performance characteristic sampling strategy, a non-interference scenario train operation observation scenario library is established, and a non-interference scenario positioning observation dataset is constructed through scenario-driven testing.

[0019] Preferably, generating a positioning observation augmented data set for a deception interference scenario based on satellite positioning deception interference features and deception interference injection testing specifically includes the following steps:

[0020] The deception jamming feature amplification based on satellite positioning forms an observation scenario library for train operation in deception jamming scenarios;

[0021] Construct an observation dataset for deception jamming scenario positioning through deception jamming injection testing;

[0022] An adversarial training strategy is used to generate observation augmentation datasets for deceptive interference scene localization.

[0023] Preferably, the combining of the positioning observation dataset and the positioning observation augmented dataset to train a general deception interference recognition model specifically includes the following steps:

[0024] Marking the data labels in the positioning observation dataset as “0” indicates that the data is real data that has not been subjected to deception interference attacks;

[0025] Marking the data labels in the positioning observation amplification data set as "1" indicates that the data is deception interference data that has been subjected to a deception interference attack;

[0026] The positioning observation data set and the positioning observation amplified data set are merged and divided into a training set and a test set through a deception interference state recognition network;

[0027] The training set and the test set are used to train a general model for deception interference recognition.

[0028] Preferably, the method further comprises the following steps:

[0029] The availability of current satellite positioning capture is judged based on the results of projection residual statistical test.

[0030] Preferably, the method further comprises the following steps:

[0031] Based on the judgment results of the deception interference identification optimization model and the virtual transponder basic data set, the capture logic of the virtual transponder is dynamically adjusted, and the capture status sequence table of the virtual transponder is updated after it is determined that the virtual transponder is successfully captured. At the moment of capture, the virtual transponder message is sent to the on-board equipment of the train operation control system.

[0032] One or more embodiments of this specification provide a virtual transponder acquisition device based on deception jamming identification, including a positioning module, a data set generation module, a model training module, and an identification module;

[0033] The positioning module is used to collect positioning observation data of the railway line to construct a positioning observation data set for an interference-free scenario;

[0034] The data set generation module is used to generate a positioning observation augmented data set of a deception interference scenario based on satellite positioning deception interference characteristics and deception interference injection tests;

[0035] The model training module is used to train a general deception interference recognition model by combining the positioning observation data set and the positioning observation amplified data set, and to iterate the model training process through an incremental training strategy of the amplified data samples to form a deception interference recognition optimization model;

[0036] The identification module is used to input the real-time acquired positioning observation data into the deception interference identification optimization model to obtain an output result, and to determine whether the current positioning data in the output result is credible based on the projection residual statistical test.

[0037] Preferably, the positioning module includes a data set construction unit, an evaluation set construction unit, a boundary feature generation unit, an observation performance discrimination unit and an observation data set construction unit;

[0038] The data set construction unit is configured to collect measurement data of a railway line to construct a line space information data set and a virtual balise basic data set.

[0039] The evaluation set construction unit is configured to collect positioning observation data of the railway line, extract observation quantities of train positioning quality after preprocessing, and construct a positioning observation quality evaluation set.

[0040] The boundary feature generation unit is configured to obtain neighborhood terrain sheltering boundary feature information of the virtual balise based on the line space information data set and the virtual balise basic data set.

[0041] The observation performance discrimination unit is configured to discriminate neighborhood navigation satellite signal observation performance of the virtual balise according to the neighborhood terrain sheltering boundary feature information of the virtual balise and the positioning observation quality evaluation set, and construct a neighborhood observation performance feature set of the virtual balise in an interference-free scenario.

[0042] The observation data set construction unit is configured to establish an interference-free scenario train operation observation scene library based on a train historical observation data set and an observation performance feature sampling strategy, and construct an interference-free scenario positioning observation data set through scene-driven testing.

[0043] One or more embodiments of the present specification provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the virtual balise capture method based on fraud interference identification as described above when executing the computer program.

[0044] One or more embodiments of the present specification provide a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the virtual balise capture method based on fraud interference identification as described above when executed by a processor.

[0045] The present invention provides a virtual transponder capture method, device, equipment and medium based on deception interference identification, which has the advantages of constructing a positioning observation data set for interference-free scenarios by collecting positioning observation data of railway lines, which can clearly present various characteristics of positioning observation data of railway lines under ideal conditions, and provide basic and real interference-free sample data for subsequent training of deception interference identification models; generating a positioning observation amplified data set for deception interference scenarios based on satellite positioning deception interference characteristics and deception interference injection tests, which can simulate a variety of deception interference scenarios that may occur in practice, and then generate corresponding positioning observation data, greatly expanding the range of interference situations covered by the data set, supplementing key interference data samples for training deception interference identification models, and helping to improve the comprehensiveness of the model in deception interference identification; combining the positioning observation data set and the positioning observation amplified data set to train a general deception interference identification model, and iterating the model through an incremental training strategy of amplified data samples. The training process forms a deception interference identification optimization model, improves the model's ability to distinguish normal and interfered positioning data, avoids misjudgment or missed judgment, and the model can continuously adjust its own parameters, structure, etc. based on new data feedback, gradually improve its adaptability to different deception interference situations, and enhance the robustness of the model; the real-time positioning observation data is input into the deception interference identification optimization model to obtain an output result, and the projection residual statistical test is used to determine whether the current positioning data in the output result is credible. It can timely identify whether the data obtained at each moment is interfered with, quickly screen out credible data, and avoid unreliable erroneous positioning data caused by deception interference from flowing into subsequent railway operation-related links, providing a reliable basis for various operations based on positioning data for the entire railway system, helping to improve the safety of train operation, the accuracy of scheduling, and the accuracy of related functions such as virtual transponder capture, ensuring that railway operations can be carried out stably and efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate one or more embodiments of this specification 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 recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0047] Figure 1 A flowchart of a virtual transponder acquisition method based on deception interference identification provided in one or more embodiments of this specification;

[0048] Figure 2 A schematic diagram of a typical layout scenario of a virtual transponder provided for one or more embodiments of this specification;

[0049] Figure 3 A schematic diagram of the online detection principle of the deception interference identification optimization model provided in one or more embodiments of this specification;

[0050] Figure 4 A flowchart of a deception interference identification optimization model provided in one or more embodiments of this specification;

[0051] Figure 5 A schematic diagram comparing the performance indicators of the deception interference identification optimization model provided in one or more embodiments of this specification;

[0052] Figure 6 A schematic diagram of a virtual transponder capture logic adjustment mechanism when subjected to a deception jamming attack, provided in one or more embodiments of this specification;

[0053] Figure 7 A schematic structural diagram of a virtual transponder capture device based on deception interference identification provided by one or more embodiments of this specification;

[0054] Figure 8 A structural diagram of a virtual transponder trusted acquisition device based on enhanced deception interference identification provided by one or more embodiments of this specification;

[0055] Figure 9 A schematic diagram of the structure of a computer device provided in one or more embodiments of this specification. DETAILED DESCRIPTION

[0056] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below in conjunction with the drawings in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this invention.

[0057] The present invention will be described in detail below with reference to specific implementation methods and the accompanying drawings.

[0058] Method Example

[0059] According to an embodiment of the present invention, a virtual transponder acquisition method based on deception interference identification is provided, such as Figure 1 FIG. 1 is a flow chart of a virtual transponder acquisition method based on deception interference identification provided by this embodiment. The virtual transponder acquisition method based on deception interference identification according to an embodiment of the present invention includes the following steps:

[0060] S110, collect positioning observation data of the railway line to construct a positioning observation data set of an interference-free scene, specifically including the following steps:

[0061] S1101, collect measurement data of the railway line to construct a line space information data set and a virtual transponder basic data set, select a target railway line to be measured, configure the signal frequency band, sampling rate, and antenna type of the receiver, after the configuration is completed, measure the target railway line data to construct the line space information data set and the virtual transponder basic data set. The line space information data set mainly stores data including the line number N(P index ) of the line, the line key point number P index , the running direction D(P index ), the three-dimensional coordinate position (L index , B index , H index ) of the line key point, wherein the subscript index represents the line key point label. The virtual transponder basic data set mainly describes the capture state data of all virtual transponders in the track line from station A to station B, including the virtual transponder number V i , the virtual transponder mileage S i , the direction flag D n , the three-dimensional coordinate position (L i , B i , H i ) of the virtual transponder, the virtual transponder capture state C(i, D n ), wherein the subscript i represents the virtual transponder serial number.

[0062] S1102, collect positioning observation data of the railway line, extract observation quantities reflecting train positioning quality after preprocessing to construct a positioning observation quality evaluation set:

[0063] K1 = [N t , (C / N0) t , θ t , ψ t , ω t ];

[0064] Wherein, N t is the number of currently visible satellites at time t, (C / N0) t is the carrier-to-noise ratio corresponding to each satellite at time t, θ t is the satellite identification number corresponding to time t, ψ t is the satellite elevation angle corresponding to each satellite at time t, and ω t is the satellite azimuth angle corresponding to each satellite at time t.

[0065] S1103 : Obtaining neighborhood terrain occlusion boundary feature information of the virtual transponder based on the line spatial information dataset and the virtual transponder basic dataset.

[0066] S1104. Determine the neighborhood navigation satellite signal observation performance of the virtual transponder based on the neighborhood terrain occlusion boundary feature information of the virtual transponder and the positioning observation quality assessment set, and construct a neighborhood observation performance feature set of the virtual transponder in an interference-free scenario.

[0067] Filter and identify satellite data based on elevation and azimuth conditions:

[0068] ψ1(V i )≤ψ t ≤ψ2(V i );

[0069] Among them, ψ1(V i ) and ψ2(V i ) represents the virtual transponder V i The lower and upper bounds of the visible elevation angle of satellites in the neighborhood are determined. Satellites that meet the requirements are screened based on the set thresholds for elevation angle, elevation angle, and carrier-to-noise ratio. The observation performance of the navigation satellite signal in the neighborhood of the current virtual transponder is determined. The detection thresholds N1 and N2 are set, and a three-class hypothesis test model is constructed:

[0070]

[0071] If the assumption H0 is established, the model output "-1" indicates that the current virtual transponder neighborhood navigation satellite signal observation performance is poor and the satellite positioning result is not credible enough. If the assumption H1 is established, the model output "0" indicates that the current virtual transponder neighborhood navigation satellite signal observation performance is average, positioning can be performed but the credibility is average. If the assumption H2 is established, the model output "1" indicates that the current virtual transponder neighborhood navigation satellite signal observation performance is good and the satellite positioning result is credible.

[0072] Based on the train historical observation dataset and the observation performance characteristic sampling strategy, a non-interference scenario train operation observation scenario library is established, and a non-interference scenario positioning observation dataset is constructed through scenario-driven testing. Among them, due to the repetitive nature of train operation, the train historical observation dataset is the satellite positioning data measured during the actual train operation.

[0073] Specifically, based on the judgment result of the neighborhood navigation satellite signal observation performance of the virtual transponder, a neighborhood observation performance feature set of the virtual transponder in the interference-free scenario is established, wherein each sample of the neighborhood observation performance feature set of the virtual transponder includes the following information items:

[0074] N t : The number of visible satellites at the current moment;

[0075] (C / N0) t : Satellite carrier-to-noise ratio corresponding to each satellite at the current moment;

[0076] The geometric dilution of precision corresponding to each satellite at the current moment;

[0077] The position precision factor of each satellite at the current moment;

[0078] The horizontal dilution of precision corresponding to each satellite at the current moment;

[0079] The vertical precision factor corresponding to each satellite at the current moment;

[0080] The performance characteristic set of the virtual transponder neighborhood observation in the non-interference scenario is composed of:

[0081]

[0082] The historical observation data set of the train on the selected target line is replayed through the train satellite positioning reproduction test. The key points of the line are analyzed in combination with the virtual balise basic data set constructed in step S1101. The typical layout scenarios of virtual balises with different navigation satellite signal observation performance on the line are selected using the observation performance feature sampling strategy, such as Figure 2 As shown, there is a schematic diagram of typical layout scenarios of the virtual transponder provided in this embodiment, including open scenes, tunnel entrance and exit scenes, mountainous scenes, urban canyon scenes, and blocked scenes under bridges, to build an interference-free scenario train operation observation scenario library.

[0083] The scene-driven test is used to build an interference-free scene positioning observation dataset. The specific test steps of the scene-driven test are as follows:

[0084] 1) Combine the line space data constructed in step S1101 and the different scene types of the train operation observation scene library, according to the number of visible satellites N at each moment t Select and configure the scene to ensure that the selected scene can cover the positioning observation data under different satellite signal conditions.

[0085] 2) Import the historical train observation dataset and perform data playback to conduct satellite positioning reproduction tests on the configured typical scenarios to simulate the satellite signal reception environment in an interference-free state.

[0086] 3) Collect data on satellite signals of selected scenes and store the data according to different scene types to ensure that the collected data has a sufficient time span and can cover various satellite observation characteristics during the train operation process.

[0087] 4) Constructing an interference-free scene positioning observation dataset based on the scene drive test results, wherein each sample in the interference-free scene positioning observation dataset includes the following information items:

[0088] t: current positioning time;

[0089] ρ t : The observed pseudorange corresponding to each satellite at the current moment;

[0090] ρ t ': The observed pseudo-range rate corresponding to each satellite at the current moment;

[0091] X t : ECEF coordinates of each satellite at the current moment;

[0092] V t : ECEF speed of each satellite at the current moment;

[0093] (C / N0) t : Satellite carrier-to-noise ratio corresponding to each satellite at the current moment;

[0094] The pseudorange after carrier phase correction corresponding to each satellite at the current moment;

[0095] The geometric dilution of precision corresponding to each satellite at the current moment;

[0096] The position precision factor of each satellite at the current moment;

[0097] The horizontal dilution of precision corresponding to each satellite at the current moment;

[0098] The vertical precision factor corresponding to each satellite at the current moment;

[0099] The positioning observation dataset that constitutes the interference-free scene is:

[0100]

[0101] S120. Generate a positioning observation augmented dataset for a deception interference scenario based on satellite positioning deception interference characteristics and deception interference injection tests.

[0102] S130. Combine the positioning observation dataset and the positioning observation amplified dataset to train a general model for deception interference identification, and iterate the model training process using an incremental training strategy of amplified data samples to form a deception interference identification optimization model, wherein the amplified data samples are satellite positioning observation datasets obtained during the historical operation of the train, and are data samples measured experimentally on the line where the train runs.

[0103] S140. During the on-the-go operation, the real-time acquired positioning observation data, i.e., the satellite positioning data of the train during the real-time operation, is input into the deception interference identification optimization model, and whether the current positioning data in the output result is credible is determined based on the projection residual statistical test.

[0104] The positioning observation data set is constructed in real time while the train is in operation, and a preliminary safety judgment is made based on the statistical test of the projection residual. First, the residual vector is calculated, and the difference between the observed data set and the theoretical value is made to form a high-dimensional residual vector r:

[0105]

[0106] Among them, z ii is the i-th observation value, is the theoretical value, r ii is the i-th component of the residual vector.

[0107] The high-dimensional residual vector is then projected into a low-dimensional statistical space to calculate the covariance matrix Σ of the residual data:

[0108]

[0109] in, is the mean vector of residuals, For the number of samples, perform eigenvalue decomposition on the covariance matrix, select the eigenvector with the larger eigenvalue as the projection basis vector, and form the projection matrix P, forming the projection matrix P = [v1, v2, ..., v k ], project the current residual vector r into the statistical space to obtain the projected low-dimensional residual vector r'=P·r, perform statistical tests on the residual vector, calculate the mean and variance of the residual after dimensionality reduction, and perform safety judgment on it:

[0110]

[0111] μ-3σ≤r′≤μ+3σ;

[0112] Among them, μ and σ are the mean and variance of the low-dimensional residual vector after projection. If r′ falls within this interval, it is judged to be normal. Otherwise, it is marked as a possible abnormality.

[0113] The three-dimensional spatial position (X(t), Y(t), Z(t)), running direction D(t), and line number N(t) of the train are calculated based on the received satellite navigation positioning observation data set. The three-dimensional spatial position is converted into coordinate position (L(t), B(t), H(t)). The virtual balise number V to be captured in this direction is determined by comparing it with the constructed virtual balise basic data set. i , extracting positioning observation data in real time and building a dynamic feature set since entering the coarse capture range of the target virtual transponder like Figure 3 , which is a schematic diagram of the online detection principle of the deception interference identification optimization model provided by this embodiment.

[0114] The method provided in this embodiment constructs a positioning observation data set for interference-free scenarios by collecting positioning observation data of railway lines, which can clearly present various features of the positioning observation data of railway lines under ideal conditions, and provide basic and real interference-free sample data for subsequent training of deception interference recognition models; generates a positioning observation amplified data set for deception interference scenarios based on satellite positioning deception interference features and deception interference injection tests, which can simulate a variety of deception interference scenarios that may occur in practice, and then generate corresponding positioning observation data, greatly expanding the range of interference situations covered by the data set, supplementing key interference data samples for training deception interference recognition models, and helping to improve the comprehensiveness of the model in deception interference recognition; combines the positioning observation data set and the positioning observation amplified data set to train a general deception interference recognition model, and iterates the model training process through an incremental training strategy of amplified data samples to form a deception interference recognition optimization model. The model improves the ability of the model to distinguish between normal and interfered positioning data, avoids misjudgment or missed judgment, and can continuously adjust its own parameters and structure according to new data feedback, gradually improves its adaptability to different deception interference situations, and enhances the robustness of the model; the real-time positioning observation data is input into the deception interference identification optimization model to obtain an output result, and the projection residual statistical test is used to determine whether the current positioning data in the output result is credible. It can timely identify whether the data obtained at each moment is interfered with, quickly screen out credible data, and avoid unreliable erroneous positioning data caused by deception interference from flowing into subsequent railway operation-related links, providing a reliable basis for various operations based on positioning data for the entire railway system, helping to improve the safety of train operation, the accuracy of scheduling, and the accuracy of related functions such as virtual transponder capture, ensuring that railway operations can be carried out stably and efficiently.

[0115] In one embodiment, step S120, generating a positioning observation augmented dataset for a deception jamming scenario based on the satellite positioning deception jamming characteristics and the deception jamming injection test, specifically includes the following steps:

[0116] S1201. A deception interference feature amplification based on satellite positioning forms an observation scenario library for train operation in deception interference scenarios. A deception interference injection tool is used to design deception interference scenarios. Considering that the implementation forms of deception interference include but are not limited to pseudo-range deception, trajectory deception, and time deception, deception commands are configured through a deception signal generation device to adjust parameters such as interference injection type, injection time, and injection power. Then, the historical observation dataset of train operation on the target line is replayed, and the configured deception interference is selected to be added to form an observation scenario library for train operation in deception interference scenarios.

[0117] S1202: Construct an observation dataset for deception interference scenario positioning through deception interference injection testing. After analyzing deception interference features, select the following feature quantities to construct the deception interference scenario positioning observation dataset:

[0118] (1) Pseudorange residual

[0119] Select the pseudorange residual between the corrected pseudorange and the observed pseudorange As a characteristic quantity, when a satellite positioning receiver is attacked by deception jamming, the corrected pseudorange changes with the pseudorange observation value, thereby affecting the pseudorange residual value, which can be used to sensitively characterize whether the receiver is affected by deception:

[0120]

[0121] Among them, ρ c,t is the corrected pseudorange, is the component of the corrected pseudorange bias due to the deception attack, which only exists in the deceived satellite, ρ a,t is the estimated pseudorange.

[0122] (2) Carrier-to-noise ratio (C / N0) t

[0123] Select carrier-to-noise ratio (C / N0) t As a characteristic quantity, when a satellite positioning receiver is attacked by deception jamming, the deception source device will force the receiver to lock on the deception signal, resulting in a stronger baseband signal, which in turn affects the carrier-to-noise ratio. Therefore, it can be used to construct identification features through changes in the carrier-to-noise ratio:

[0124]

[0125] in, and are the carrier power and noise power in the deceptive scenario, respectively. Generally speaking, (C / N0) t The higher the value, the better the communication quality.

[0126] (3) Carrier smoothed pseudorange residual λs,t

[0127] Select carrier smoothed pseudorange residual λ s,t As a characteristic quantity, when a satellite positioning receiver is attacked by deception jamming, the high precision of the carrier phase makes it very sensitive to the deception jamming, which in turn affects the change of the carrier smoothed pseudorange. It can be used to sensitively characterize whether the receiver is affected by deception:

[0128]

[0129] Among them, ρ s,t is the carrier smoothed pseudorange at time t, φ t is the carrier phase at time t, M is the smoothing time constant, λ is the wavelength of the corresponding frequency, is the carrier smoothed pseudorange equivalent deviation component caused by the deception attack, and the carrier smoothed pseudorange residual can be obtained:

[0130]

[0131] Then the positioning observation dataset for constructing the deception interference scenario is:

[0132]

[0133] S1203. Use the adversarial training strategy to generate an augmented observation dataset for deceptive interference scene positioning. After obtaining the deceptive interference scene positioning observation dataset, use the deceptive scene sample generation network to perform sample enhancement. The deceptive scene sample generation network consists of two neural networks: a generator and a discriminator. The objective function of the adversarial network is:

[0134] min G max D E x～pdata [logD(x)]+E z～pz(z) [log(1-D(G(z)))];

[0135] Among them, min G max D Represents the game relationship between the generator G and the discriminator D, E x～pdata [logD(x)] represents the discriminator’s prediction of the real sample, E z～pz(z) [log(1-D(G(z)))] represents the prediction part of the discriminator for the generated sample, p data is the distribution of real data, p z(z) is the distribution of the generator input noise. The generator and the discriminator are jointly optimized through game theory, which ultimately enables the generator to generate deceptive interference samples that are similar to the real data distribution, thereby forming a deceptive interference scene positioning observation amplification dataset.

[0136] The method provided in this embodiment can effectively integrate satellite positioning deception interference features and injection testing methods. On the one hand, based on the deception interference features, a train operation observation scenario library is expanded to accurately simulate complex interference scenarios. On the other hand, with the help of deception interference injection testing, a positioning observation dataset is constructed to provide diverse materials for analysis. Finally, an adversarial training strategy is used to generate an observation amplification dataset, which comprehensively improves the ability to recognize, analyze and respond to satellite positioning deception interference scenarios, and enhances the system's positioning reliability and stability in complex interference environments.

[0137] In one embodiment, step S130, combining the positioning observation dataset K true and the positioning observation amplification dataset K sp Training a general model for deception interference recognition, such as Figure 4 FIG. 5 is a flow chart of the deception interference identification optimization model provided in this embodiment, which specifically includes the following steps:

[0138] S1301, the positioning observation data set K true The data label in is marked as “0”, which means that the data is real data that has not been attacked by deception interference.

[0139] S1302: amplify the positioning observation dataset K sp The data label marked as "1" in indicates that the data is deceptive jamming data that has been attacked by deceptive jamming.

[0140] S1303, select a deception interference state recognition network to perform deception interference recognition and classification, and use the deception interference state recognition network to classify the positioning observation data set K true The dataset K is augmented with the positioning observations sp After merging, the data is divided into a training set and a test set. The data with a preset ratio of 1 / Ω is used as the training set of the model, and the remaining data with a ratio of (1-1 / Ω) is used as the test set. In this embodiment, if Ω=1.25, 80% of the data is used as the training set of the model and 20% of the data is used as the test set to complete the training of the deception interference state recognition network. The model objective function is:

[0141]

[0142] in, I j1 is the set of instances in leaf j1, with the optimal weight g iand h i are the first-order and second-order partial derivatives of the loss function, respectively. and γ are regularization parameters, T is the number of leaves, is the weight of the leaf node. The general model for deception interference recognition uses a decision tree model. The basis for selecting feature splitting points is to maximize gain (Gain). All features and all possible splitting points are traversed:

[0143]

[0144] The update of the predicted value is calculated by the weight of each leaf node, and the optimal weight for:

[0145]

[0146] Substituting the weights into the objective function, we can get the minimized loss value of the leaf node:

[0147]

[0148] Among them, g i and h i are the first-order and second-order partial derivatives of the loss function, respectively. is the instance set in leaf j1, IL and IR are the sample index sets of the left and right child nodes respectively, and I is the sample index set of the parent node. and γ are regularization parameters, T is the number of leaves, is the weight of the leaf node.

[0149] Based on the trained initial general model for deception interference recognition, amplified data of deception interference scene positioning observations are introduced in stages, and training is performed using increasing sample sizes and gradually deepening interference features. After each stage of training, the precision rate p and the test threshold ε are set as the criteria for judging the model training status. When the model's precision rate p satisfies p>ε, it is confirmed that the model has met the training requirements of this stage and can enter the next stage of training; otherwise, the model needs to continue to iterate on the data set of this stage until the precision rate can meet the set test threshold. Through this incremental training strategy of amplified data samples, the model training process is iterated to form an optimized model for deception interference recognition.

[0150] Based on the trained initial deception jamming recognition general model, the Bayesian optimization algorithm is used to find the optimal hyperparameter combination and optimize the model:

[0151] 1) Initialization: Randomly generate n hyperparameters as the initial hyperparameter combination, calculate the accuracy of the above model output, and use it as the evaluation indicator of the Bayesian optimization algorithm. The objective function is defined as:

[0152] f(x) = Accuracy(x);

[0153] Where Accuracy(x) represents the accuracy on the validation set after using the hyperparameter combination x to train the general model for deception interference recognition.

[0154] 2) Constructing a Gaussian process model: Use the existing hyperparameter precision to fit the Gaussian process. μ(x) is the predicted mean, which represents the estimated value of the objective function at point x. k(x, x′) is the kernel function used to quantify the correlation between x and x′:

[0155] f(x)~GP(μ(x),k(x,x′));

[0156] 3) Optimize the sampling function: Select Expected Improvement (EI) as the sampling function to determine the next hyperparameter combination to be evaluated. The larger the EI value, the higher the exploration value of the hyperparameter combination:

[0157] EI(x)=(μ(x)-f best )Φ(z)+σ(x)φ(z);

[0158] x next =argmax x∈X EI(x);

[0159] where f best is the current optimal objective function value, μ(x) is the predicted mean of the objective function by the Gaussian process, σ(x) is the predicted standard deviation of the Gaussian process, Φ(z) and φ(z) are the cumulative distribution function and probability density function of the standard normal distribution, respectively.

[0160] 4) Update hyperparameter combination: For the selected hyperparameter combination, train the deception interference recognition general model and calculate the objective function value, and add the new evaluation results to the sampling set middle: Then refit the Gaussian process from Select the hyperparameter combination with the best objective function value

[0161] 5) Check whether the algorithm iteration termination conditions are met. If so, end the optimization and output the global optimal hyperparameter combination as the model hyperparameters; otherwise, return to step 3 and continue the optimization.

[0162] The observation augmentation data in the phased introduction of deception jamming scenario is positioned, the incremental sample size and the gradually deepened interference characteristics are used for training, and the precision p and the test threshold ε are set as the standards for judging the model training condition. After each stage of training is completed, the model performance is evaluated. When the precision p of the model meets p>ε, it is confirmed that the model has met the training requirements of this stage, and can enter the next stage of training. Otherwise, the model needs to continue iteration on the data set of this stage until the precision can meet the set test threshold. Through the incremental training strategy of augmented data sample iteration model training process, the deception jamming recognition optimization model is formed.

[0163] The performance index result comparison diagram is used to test the deception jamming recognition optimization model performance by using non-modeling samples. The non-modeling samples are selected from the non-interference and deception jamming data samples that are not used in model training, as the model validation data set. For the output results of the model, the discrimination degree of the model to the real samples and the deception samples is tested by the harmonic ratio:

[0164]

[0165] Wherein, P real is the prediction accuracy of the model to the real samples, P fake is the prediction accuracy of the model to the deception jamming samples, the test threshold ξ is set, and the test performance of the deception jamming recognition optimization model in the complex scene is evaluated by analyzing the harmonic ratio distribution of the model on the validation data set:

[0166] 1) When the harmonic ratio H>ξ, or even the harmonic ratio is close to 1, the performance of the deception jamming recognition optimization model is good, and the model can perform subsequent security discrimination work;

[0167] 2) When the harmonic ratio H<ξ, or even the harmonic ratio is close to 0, the performance of the deception jamming recognition optimization model is poor, and the model needs to be retrained, and then the model is tested by the harmonic ratio.

[0168] As Figure 5 shown, it is the performance index result comparison diagram of the deception jamming recognition optimization model provided in the embodiment.

[0169] S1304, train the deception jamming recognition general model by using the training set and the test set.

[0170] The method provided in this embodiment trains a general model for identifying spoofing interference by cleverly combining a positioning observation dataset with an augmented positioning observation dataset. First, it accurately distinguishes between two types of data labels, visually presenting the data's nature. Then, using a spoofing interference state recognition network, it rationally divides the training and test sets, fully integrating data resources with different characteristics. Finally, using these sets to train the model effectively improves the model's ability to accurately identify satellite positioning spoofing interference, enabling it to quickly and accurately determine whether data has been subjected to spoofing interference, thereby enhancing the system's ability to defend against spoofing interference.

[0171] In one embodiment, the following steps are further included:

[0172] The availability of current satellite positioning capture is judged based on the results of projection residual statistical test.

[0173] Dynamic feature sets built in real time Using the Min-Max principle, normalize each feature quantity contained in the current feature set and map each feature quantity to [0,1]:

[0174]

[0175] where x i is the original value of the feature quantity, x min and x max are the minimum and maximum values ​​of the feature quantities respectively. The normalized dynamic feature set is:

[0176]

[0177] The offline trained deception jamming identification optimization model is called online for detection, and enhanced security discrimination is implemented. A binary hypothesis testing model is established based on the discrimination results:

[0178]

[0179] If the assumption Y0 is established, the model output "0" indicates that the system is not under deception attack at the current moment. At this time, the virtual transponder can rely on the satellite positioning result P gnss Normal capture; if the assumption Y1 is established, the model output "1" indicates that the system is under a spoofing attack at the current moment. At this time, the satellite positioning result is unavailable, and the virtual transponder needs to use the non-satellite navigation positioning result P non-gnss to capture.

[0180] The method provided in this embodiment, using projection residual statistical test results, can accurately and quickly determine the current availability of satellite positioning capture. This scientific and quantitative testing method effectively identifies the quality of satellite positioning signals and promptly identifies potential problems, providing a reliable basis for the stable operation of subsequent navigation and positioning-related systems. It avoids a series of deviations caused by unavailable positioning capture and ensures the efficient and accurate operation of services that rely on satellite positioning.

[0181] In one embodiment, the following steps are further included:

[0182] Based on the judgment results of the deception interference identification optimization model and the virtual transponder basic data set, the capture logic of the virtual transponder is dynamically adjusted, and the capture status sequence table of the virtual transponder is updated after it is determined that the virtual transponder is successfully captured. At the moment of capture, the virtual transponder message is sent to the on-board equipment of the train operation control system.

[0183] The virtual transponder positioning processing logic is adjusted based on the output of the deception interference identification optimization model. Based on the deception interference detection judgment results, confirmation is carried out in two cases:

[0184] 1) If the output of the deception interference identification optimization model is "0", it means that the system is not under deception attack at the current moment, and the satellite positioning result is judged to be "available state". At this time, it is a normal state, and the virtual transponder relies normally on the satellite navigation positioning result P gnss To locate,

[0185] 2) If the output of the deception interference identification optimization model is "1", it indicates that the system is under deception attack at the current moment, and the satellite positioning result is judged to be "unavailable". At this time, the virtual transponder needs to rely on the positioning result of the non-satellite navigation positioning sensor P non-gnss to locate.

[0186] like Figure 6 The figure shows the schematic diagram of the virtual transponder capture logic adjustment mechanism when it is attacked by deception interference provided by this embodiment. Among them, R is the set capture radius, H represents the distance between the actual position of the train and the reference point of the virtual transponder, h represents the distance between the position of the train under deception attack and the virtual transponder, and h' represents the positioning result P using the non-satellite navigation positioning sensor. non-gnss The calculated distance to the virtual transponder.

[0187] Based on the above satellite navigation positioning results P gnss Compared with the positioning result of non-satellite navigation positioning sensor P non-gnssImplement the capture judgment of the virtual balise, extract the line space information data in the basic data set of the virtual balise according to the current running data of the train, and calculate the current running mileage S(t) of the train. According to the current running mileage of the train and the corresponding running mileage of the previous positioning cycle, calculate the running speed of the train along the track V(t) = |S(t)-S(t-τ)| / τ, and predict the train position at the future time based on this. τ is the preset calculation cycle length, k = 1, 2, ..., n, where n is the maximum detection step length.

[0188]

[0189] Among them, C cap is the capture radius of the virtual balise. When the forward detection performed at time t according to the maximum detection step n satisfies the above formula, it is determined that the train has entered the capture radius of the target virtual balise, and the virtual balise capture can be completed.

[0190] After the target virtual transponder is captured, the virtual transponder capture state sequence table is updated, and the three-dimensional coordinate position (L i ,B i ,H i ), sent to the train position calculation module, modifying the capture state C of the current target virtual balise i If it is captured, the virtual balise state sequence in the virtual balise basic database is updated, and the virtual balise closest to the current target virtual balise along the actual running direction of the train is selected as the target virtual balise to be captured in the subsequent operation process of the system, and the number of the virtual balise to be captured is updated to V i+1 .

[0191] The method provided in this embodiment dynamically optimizes the virtual transponder capture logic by combining the results of the deception interference identification optimization model with the basic virtual transponder dataset. It accurately determines the capture status and timely updates the capture status sequence table. Once the capture moment is confirmed, the virtual transponder message is immediately sent to the onboard equipment of the train operation control system. This ensures the timeliness and accuracy of information received by the train, effectively improving the accuracy and stability of train operation control, reducing the risk of information errors caused by interference, and ensuring safe and efficient train operation.

[0192] Device embodiment

[0193] According to an embodiment of the present invention, a virtual transponder capture device based on deception interference identification is provided, such as Figure 7As shown, this is a structural diagram of the virtual transponder capture device based on deception interference identification provided in this embodiment. The virtual transponder capture device based on deception interference identification according to an embodiment of the present invention includes a positioning module 71, a data set generation module 72, a model training module 73 and an identification module 74.

[0194] The positioning module 71 is used to collect the positioning observation data of the railway line to construct a positioning observation data set for an interference-free scenario.

[0195] The data set generation module 72 is configured to generate a positioning observation augmented data set for a deception jamming scenario based on the satellite positioning deception jamming characteristics and the deception jamming injection test.

[0196] The model training module 73 is used to train a general deception interference recognition model by combining the positioning observation dataset and the positioning observation amplified dataset, and to form a deception interference recognition optimization model by iterating the model training process through an incremental training strategy of amplified data samples.

[0197] The identification module 74 is used to input the real-time acquired positioning observation data into the deception interference identification optimization model to obtain an output result, and determine whether the current positioning data in the output result is credible based on the projection residual statistical test.

[0198] The device provided in this embodiment comprises a positioning module 71 which constructs a positioning observation data set of an interference-free scenario by collecting positioning observation data of a railway line, and can clearly present various features of the positioning observation data of the railway line under an ideal state, and provides basic and real interference-free sample data for the subsequent training of a deception interference recognition model; a data set generation module 72 generates a positioning observation amplified data set of a deception interference scenario based on satellite positioning deception interference features and a deception interference injection test, and can simulate a variety of deception interference scenarios that may occur in practice, and then generate corresponding positioning observation data, which greatly expands the range of interference situations covered by the data set, and supplements key interference data samples for training the deception interference recognition model, and helps to improve the comprehensiveness of the model in identifying deception interference; a model training module 73 trains a general deception interference recognition model in combination with the positioning observation data set and the positioning observation amplified data set, and forms a deception interference recognition model by iterating the model training process through an incremental training strategy of amplified data samples. The deception interference identification optimization model improves the model's ability to distinguish normal and interfered positioning data, avoids misjudgment or missed judgment, and the model can continuously adjust its own parameters and structure based on new data feedback, gradually improve its adaptability to different deception interference situations, and enhance the robustness of the model; the identification module 74 inputs the real-time positioning observation data into the deception interference identification optimization model to obtain an output result, and judges whether the current positioning data in the output result is credible based on the projection residual statistical test. It can timely identify whether the data obtained at each moment is interfered with, quickly screen out credible data, and avoid unreliable erroneous positioning data caused by deception interference from flowing into subsequent railway operation-related links, providing a reliable basis for various operations based on positioning data for the entire railway system, helping to improve the safety of train operation, the accuracy of scheduling, and the accuracy of related functions such as virtual transponder capture, ensuring that railway operations can be carried out stably and efficiently.

[0199] In one embodiment, Figure 8 As shown, the virtual transponder trusted capture device based on enhanced deception interference identification provided by this embodiment includes:

[0200] The real-time positioning module is used to receive the original observation information of satellite positioning receivers and other non-satellite navigation auxiliary sensors in real time, complete the spatiotemporal calibration and format conversion of the original data of different sensors, and provide positioning observation data in a standard format to the acquisition and generation module.

[0201] The deception interference identification module is used to extract the data set constructed by the real-time positioning module, and integrate the interference-free scene positioning observation data set and the deception interference scene positioning observation amplified data set to complete the training of the general deception interference identification model. The deception interference identification optimization model is trained through the iterative model training process of the incremental training strategy of the amplified data samples.

[0202] The preliminary safety judgment module is used to make preliminary judgments on the data set acquired in real time. It uses the projection residual statistical test technology to quickly screen abnormal data and determine whether the current positioning data is credible, providing preliminary safety filtering for subsequent enhanced safety judgment.

[0203] The enhanced security judgment module is used to extract the judgment results of the deception interference recognition model module and the preliminary security judgment module, and further judge the current satellite positioning capture availability status based on the optimized deception interference recognition model.

[0204] The capture operation module is used to extract the judgment results of the enhanced safety judgment module and the virtual transponder basic data set of the acquisition generation module. According to the real-time positioning and safety judgment results, it dynamically adjusts the virtual transponder capture logic, and updates the virtual transponder capture status sequence table after determining that the target virtual transponder is successfully captured. At the same time, at the moment of capture, the virtual transponder message is sent to the on-board equipment of the train operation control system.

[0205] The device provided in this embodiment realizes the detection of deception interference signals during the capture process of the virtual transponder, judges the current satellite positioning capture availability status based on the deception interference identification optimization model, and optimizes the capture logic of the virtual transponder according to the judgment result, providing reliable guarantee for the normal operation of the train and effectively supporting the promotion and application of virtual transponders.

[0206] In one embodiment, the positioning module 71 includes a data set construction unit 7101 , an evaluation set construction unit 7102 , a boundary feature generation unit 7103 , an observation performance determination unit 7104 and an observation data set construction unit 7105 .

[0207] The data set construction unit 7101 is used to collect the measurement data of the railway line to construct the line spatial information data set and the virtual balise basic data set.

[0208] The evaluation set construction unit 7102 is used to collect the positioning observation data of the railway line, extract the observation quantity of the train positioning quality after preprocessing, and construct the positioning observation quality evaluation set.

[0209] The boundary feature generating unit 7103 is configured to obtain the neighborhood terrain occlusion boundary feature information of the virtual transponder based on the line spatial information dataset and the virtual transponder basic dataset.

[0210] The observation performance judgment unit 7104 is used to judge the neighborhood navigation satellite signal observation performance of the virtual transponder based on the neighborhood terrain occlusion boundary feature information of the virtual transponder and the positioning observation quality assessment set, and construct a neighborhood observation performance feature set of the virtual transponder in an interference-free scenario.

[0211] The observation data set construction unit 7105 establishes a non-interference scenario train operation observation scenario library based on the train historical observation data set and the observation performance characteristic sampling strategy, and constructs a non-interference scenario positioning observation data set through scenario-driven testing.

[0212] The device provided in this embodiment realizes comprehensive and accurate data utilization and performance discrimination through the collaborative operation of multiple units. The data set construction unit collects data to form a basic data set of lines and transponders, laying a solid foundation for subsequent analysis; the evaluation set construction unit preprocesses the positioning observation data and outputs a positioning observation quality assessment set to facilitate accurate evaluation; the boundary feature generation unit mines key boundary feature information; the observation performance discrimination unit fuses multi-source information to discriminate the transponder neighborhood observation performance and constructs a feature set; the observation data set construction unit integrates historical data and sampling strategies to create an observation scenario library and positioning observation data set, providing the railway operation system with all-round support from basic data storage, quality control to precise positioning observation, ensuring accurate and reliable train operation navigation, and improving operation efficiency and safety.

[0213] The embodiment of the present invention is an apparatus embodiment corresponding to the above-mentioned method embodiment. The specific operations of the processing steps of each module can be understood by referring to the description of the method embodiment, and will not be repeated here.

[0214] like Figure 9 As shown, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the virtual transponder capture method based on deception interference identification in the above-mentioned embodiment when the computer program is executed by a processor, or implements the virtual transponder capture method based on deception interference identification in the above-mentioned embodiment when the computer program is executed by a processor.

[0215] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0216] Each embodiment in this specification is described in a progressive manner. The same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the device or system embodiments, since they are basically similar to the method embodiments, the description is relatively simple. For the relevant parts, refer to the partial description of the method embodiments. The device and system embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. A person of ordinary skill in the art can understand and implement it without making any creative efforts.

[0217] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and the contents not described in detail in the specification of the present invention are common knowledge to those skilled in the art.

Claims

1. A virtual transponder acquisition method based on deception interference identification, characterized in that: The following steps are involved: Collecting the positioning observation data of the railway line to build a positioning observation dataset for the interference-free scene includes the following steps: Collect railway line measurement data to build line spatial information dataset and virtual balise basic dataset; Collect railway line positioning observation data, extract train positioning quality observations after preprocessing, and construct a positioning observation quality assessment set; Obtaining neighborhood terrain occlusion boundary feature information of the virtual transponder based on the line spatial information dataset and the virtual transponder basic dataset; Determine the neighborhood navigation satellite signal observation performance of the virtual transponder based on the neighborhood terrain occlusion boundary feature information of the virtual transponder and the positioning observation quality assessment set, and construct a neighborhood observation performance feature set of the virtual transponder in an interference-free scenario; Based on the historical train observation dataset and the observation performance characteristic sampling strategy, a non-interference scenario train operation observation scenario library is established, and a non-interference scenario positioning observation dataset is constructed through scenario-driven testing; Based on the satellite positioning deception jamming characteristics and deception jamming injection test, a positioning observation augmented dataset for the deception jamming scenario is generated, specifically including the following steps: The deception jamming feature amplification based on satellite positioning forms an observation scenario library for train operation in deception jamming scenarios; Construct an observation dataset for deception jamming scenario positioning through deception jamming injection testing; Use adversarial training strategies to generate observation augmentation datasets for deceptive interference scene localization; Training a general deception interference recognition model by combining the positioning observation data set and the positioning observation amplified data set, and forming a deception interference recognition optimization model by iterating the model training process through an incremental training strategy of the amplified data samples; The real-time acquired positioning observation data is input into the deception interference identification optimization model to obtain an output result, and whether the current positioning data in the output result is credible is judged based on the projection residual statistical test, including: calculating the residual vector, subtracting the observation data set from the theoretical value to form a high-dimensional residual vector, projecting the high-dimensional residual vector to a low-dimensional statistical space, calculating the covariance matrix of the residual data, performing eigenvalue decomposition on the covariance matrix, selecting the eigenvector with a larger eigenvalue as the projection basis vector to form a projection matrix, projecting the current residual vector to the statistical space to obtain a projected low-dimensional residual vector, performing a statistical test on the residual vector, calculating the mean and variance of the residual after dimensionality reduction, and performing a security judgment on it.

2. The virtual transponder acquisition method based on deception interference identification according to claim 1 is characterized in that: The method of combining the positioning observation dataset and the positioning observation augmented dataset to train a general deception interference recognition model specifically includes the following steps: Marking the data labels in the positioning observation data set as "0" indicates that the data is real data that has not been subjected to deception interference attacks; Marking the data labels in the positioning observation amplification data set as "1" indicates that the data is deception interference data that has been attacked by deception interference; The positioning observation data set and the positioning observation amplified data set are merged and divided into a training set and a test set through a deception interference state recognition network; The training set and the test set are used to train a general model for deception interference recognition.

3. The virtual transponder acquisition method based on deception interference identification according to claim 1, characterized in that: The following steps are also included: The availability of current satellite positioning capture is judged based on the results of projection residual statistical test.

4. The virtual transponder acquisition method based on deception interference identification according to claim 1, characterized in that: The following steps are also included: Based on the judgment results of the deception interference identification optimization model and the virtual transponder basic data set, the capture logic of the virtual transponder is dynamically adjusted, and the capture status sequence table of the virtual transponder is updated after it is determined that the virtual transponder is successfully captured. At the moment of capture, the virtual transponder message is sent to the on-board equipment of the train operation control system.

5. A virtual transponder capture device based on deception interference identification, characterized in that: Includes positioning module, data set generation module, model training module and recognition module; The positioning module is used to collect the positioning observation data of the railway line to construct a positioning observation data set for the interference-free scene. The positioning module includes a data set construction unit, an evaluation set construction unit, a boundary feature generation unit, an observation performance discrimination unit and an observation data set construction unit; The data set construction unit is used to collect measurement data of the railway line to construct a line spatial information data set and a virtual balise basic data set; The evaluation set construction unit is used to collect the positioning observation data of the railway line, extract the observation value of the train positioning quality after pre-processing, and construct the positioning observation quality evaluation set; The boundary feature generating unit is configured to obtain the neighborhood terrain occlusion boundary feature information of the virtual transponder based on the line space information dataset and the virtual transponder basic dataset; The observation performance determination unit is configured to determine the neighborhood navigation satellite signal observation performance of the virtual transponder based on the neighborhood terrain occlusion boundary feature information of the virtual transponder and the positioning observation quality assessment set, and to construct a neighborhood observation performance feature set of the virtual transponder in an interference-free scenario; The observation data set construction unit establishes a non-interference scenario train operation observation scenario library based on the train historical observation data set and the observation performance characteristic sampling strategy, and constructs a non-interference scenario positioning observation data set through scenario-driven testing; The data set generation module is used to generate a positioning observation augmented data set for a deception interference scenario based on satellite positioning deception interference characteristics and deception interference injection tests. The specific configuration is as follows: The deception jamming feature amplification based on satellite positioning forms an observation scenario library for train operation in deception jamming scenarios; Construct an observation dataset for deception jamming scenario positioning through deception jamming injection testing; Use adversarial training strategies to generate observation augmentation datasets for deceptive interference scene localization; The model training module is used to train a general deception interference recognition model by combining the positioning observation data set and the positioning observation amplified data set, and to iterate the model training process through an incremental training strategy of the amplified data samples to form a deception interference recognition optimization model; The identification module is used to input the real-time acquired positioning observation data into the deception interference identification optimization model to obtain an output result, and judge whether the current positioning data in the output result is credible based on the projection residual statistical test, including: calculating the residual vector, subtracting the observation data set from the theoretical value to form a high-dimensional residual vector, projecting the high-dimensional residual vector into a low-dimensional statistical space, calculating the covariance matrix of the residual data, performing eigenvalue decomposition on the covariance matrix, selecting the eigenvector with the larger eigenvalue as the projection basis vector to form a projection matrix, projecting the current residual vector into the statistical space to obtain the projected low-dimensional residual vector, performing a statistical test on the residual vector, calculating the mean and variance of the residual after dimensionality reduction, and performing a security judgment on it.

6. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the virtual transponder acquisition method for deception interference identification according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the virtual transponder acquisition method for deceptive interference identification according to any one of claims 1 to 4 are implemented.

Images