Geomagnetic navigation method and system based on vector data

Abstract

The present application relates to the field of geomagnetic navigation technology, and specifically discloses a geomagnetic navigation method and system based on vector data, which analyzes the magnetic observation matrix and calibrated magnetic observation data by using data processing and analysis algorithms based on artificial intelligence and deep learning to extract the spatial and temporal characteristics of magnetic observation semantics, and estimates the unknown parameters of the linear magnetic observation model based on the cross-modal joint interactive coding representation between the two. In this way, the unknown parameters of the linear magnetic observation model can be automatically inferred based on the cross-modal joint semantics between the spatial and temporal characteristics of the magnetic observation data, thereby avoiding the inaccuracy of the traditional method of estimating unknown parameters and fitting the linear magnetic observation model. In this way, the interfering magnetic field generated by the aircraft can be more effectively separated, thereby improving the accuracy and reliability of geomagnetic navigation and positioning.

Description

Technical Field

[0001] The present application relates to the field of geomagnetic navigation technology, and more specifically, to a geomagnetic navigation method and system based on vector data. Background Art

[0002] Geomagnetic navigation technology uses the characteristics of the Earth's magnetic field to determine the position of an aircraft by matching the collected geomagnetic information with a pre-built geomagnetic field database. This technology has been widely used in many fields due to its advantages such as no long-term accumulated errors and not being easily interfered with. However, when an aircraft equipped with a magnetometer uses geomagnetic navigation, it will be interfered by the magnetic field generated by its own ferromagnetic materials and electronic devices, resulting in errors in the collected geomagnetic data, which in turn affects the accuracy of positioning and navigation.

[0003] In response to the above technical problems, Chinese patent CN117213472A proposes a geomagnetic navigation method, device and electronic device based on vector data, which can obtain the vector geomagnetic observation data collected by the aircraft equipped with geomagnetic measurement sensors when flying in the calibration area, determine the magnetic change rate of the vector geomagnetic observation data in the specified direction, and then determine the linear magnetic observation model containing unknown parameters representing the relationship between the vector geomagnetic observation data collected by the aircraft and the interference data generated by the aircraft and solve the unknown parameters. This allows the aircraft to solve the interference magnetic data generated by the aircraft through the collected vector geomagnetic observation data and the linear magnetic observation model that solves the unknown parameters when flying in the target area, and then eliminate the influence of the interference magnetic data in the geomagnetic observation data collected by the aircraft, and obtain the actual magnetic data of the aircraft's position, so as to improve the accuracy of navigation.

[0004] However, in the above-mentioned geomagnetic navigation method, the process of determining unknown parameters is to fit the linear magnetic observation model by the magnetic change rate corresponding to each time point and the calibration magnetic observation data. In this way, the accuracy of the unknown parameters is highly dependent on the quality of the calibration magnetic observation data at a single time point. If there is noise or error in the collected data, or if there are missing values ​​or outliers in the data collection process, it will directly affect the construction of the magnetic observation matrix and the fitting results of the linear magnetic observation model.

[0005] Therefore, an optimized geomagnetic navigation scheme based on vector data is desired. Summary of the invention

[0006] The present application provides a geomagnetic navigation method and system based on vector data, which can automatically infer the unknown parameters of the linear magnetic observation model based on the cross-modal joint semantics between the spatial characteristics and temporal characteristics of the magnetic observation data, thereby avoiding the inaccuracy of the traditional method of estimating the unknown parameters and fitting the linear magnetic observation model. In this way, the interfering magnetic field generated by the aircraft can be more effectively separated, thereby improving the accuracy and reliability of geomagnetic navigation and positioning.

[0007] In a first aspect, a geomagnetic navigation method based on vector data is provided, comprising: obtaining vector geomagnetic observation data collected when an aircraft equipped with a geomagnetic measurement sensor is flying, as calibration magnetic observation data; determining the magnetic change rate of the calibration magnetic observation data in a specified direction; determining unknown parameters of a linear magnetic observation model based on the calibration magnetic observation data and the magnetic change rate, the parameters being used to represent the relationship between the geomagnetic observation data collected by the aircraft and the interference magnetic data generated by the aircraft; obtaining vector geomagnetic observation data used when the aircraft is flying in a target area, as target observation data; determining the interference magnetic data generated by the aircraft based on the linear magnetic observation model for solving the unknown parameters and the target observation data; determining the actual magnetic data of the position of the aircraft when flying in the target area based on the target observation data and the interference magnetic data; navigating the flight of the aircraft in the target area based on the actual magnetic data;

[0008] Wherein, based on the calibrated magnetic observation data and the magnetic change rate, determining the unknown parameters of the linear magnetic observation model includes: determining the magnetic observation matrix according to the magnetic change rate and the component magnetic observation data of the calibrated magnetic observation data in the specified direction; determining the unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data;

[0009] Wherein, determining the unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data includes:

[0010] Performing feature extraction on the magnetic observation matrix to obtain magnetic observation semantic coding features;

[0011] Performing time series encoding on the calibration magnetic observation data to obtain time series encoding features of the magnetic observation data;

[0012] Performing cross-modal continuous prompt joint encoding on the magnetic observation semantic encoding features and the magnetic observation data temporal encoding features to obtain magnetic observation spatiotemporal cross-modal joint encoding features;

[0013] Based on the spatiotemporal cross-modal joint encoding features of the magnetic observation, the unknown parameters of the linear magnetic observation model are determined.

[0014] In a second aspect, a geomagnetic navigation system based on vector data is provided, comprising: a magnetic observation matrix feature extraction module, used for extracting features from the magnetic observation matrix to obtain magnetic observation semantic coding features;

[0015] A calibration magnetic observation data time series encoding module, used for performing time series encoding on the calibration magnetic observation data to obtain a time series encoding feature of the magnetic observation data;

[0016] A cross-modal continuous prompt joint encoding module, used for performing cross-modal continuous prompt joint encoding on the magnetic observation semantic encoding feature and the magnetic observation data temporal encoding feature to obtain a magnetic observation spatiotemporal cross-modal joint encoding feature;

[0017] The unknown parameter determination module is used to determine the unknown parameters of the linear magnetic observation model based on the spatiotemporal cross-modal joint encoding characteristics of the magnetic observation.

[0018] The present application provides a geomagnetic navigation method and system based on vector data, which analyzes the magnetic observation matrix and calibrated magnetic observation data by adopting a data processing and analysis algorithm based on artificial intelligence and deep learning to extract the spatial and temporal characteristics of the magnetic observation semantics, and estimates the unknown parameters of the linear magnetic observation model based on the cross-modal joint interactive coding representation between the two. In this way, the unknown parameters of the linear magnetic observation model can be automatically inferred based on the cross-modal joint semantics between the spatial and temporal characteristics of the magnetic observation data, thereby avoiding the inaccuracy of the traditional method of estimating unknown parameters and fitting the linear magnetic observation model. In this way, the interfering magnetic field generated by the aircraft can be more effectively separated, thereby improving the accuracy and reliability of geomagnetic navigation and positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0020] Figure 1 This is a schematic flowchart of a geomagnetic navigation method based on vector data according to an embodiment of the present application.

[0021] Figure 2 A schematic flowchart of determining unknown parameters of a linear magnetic observation model based on the calibrated magnetic observation data and the magnetic change rate in a geomagnetic navigation method based on vector data in an embodiment of the present application.

[0022] Figure 3A schematic flowchart of determining unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data in the geomagnetic navigation method based on vector data in an embodiment of the present application.

[0023] Figure 4 In the geomagnetic navigation method based on vector data in an embodiment of the present application, a data flow diagram of determining unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data.

[0024] Figure 5 A schematic block diagram of a geomagnetic navigation system based on vector data according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0026] It should be understood that the scheme in the existing patent mainly focuses on the data at a single time point when processing the calibration magnetic observation data and fitting the linear magnetic observation model to determine the unknown parameters, while ignoring the temporal correlation between different time points. This processing method may result in the inability to fully utilize the temporal dynamic information in the data, thereby affecting the accuracy and robustness of the model. In other words, the geomagnetic observation data will change with time and position during the flight of the aircraft. If the temporal characteristics of these changes cannot be effectively captured, the model may perform poorly when processing dynamic scenes, especially when the aircraft performs complex maneuvers. In addition, the scheme in the existing patent mainly relies on the direct fitting of a simple magnetic observation matrix and calibration magnetic observation data when determining the unknown parameters of the linear magnetic observation model. Although this method can preliminarily determine the unknown parameters, it lacks the mining of the deep-level interactive features between the magnetic observation matrix and the calibration magnetic observation data, which may result in low generalization ability and prediction accuracy of the model. This is because in practical applications, the relationship between geomagnetic observation data and the magnetic observation matrix may be very complex, and simple fitting may not be able to fully capture these complex relationships. Therefore, when the model faces new or unseen data, it may have large errors, affecting the accuracy of navigation.

[0027] In view of the above technical problems, in the technical solution of this application, a geomagnetic navigation method based on vector data is proposed. Figure 1 FIG. 1 is a schematic flow chart of a geomagnetic navigation method based on vector data according to an embodiment of the present application. Figure 1As shown, the geomagnetic navigation method based on vector data includes: S1, obtaining vector geomagnetic observation data collected when an aircraft equipped with a geomagnetic measurement sensor is flying, as calibration magnetic observation data; S2, determining the magnetic change rate of the calibration magnetic observation data in a specified direction; S3, based on the calibration magnetic observation data and the magnetic change rate, determining the unknown parameters of the linear magnetic observation model, the parameters are used to represent the relationship between the geomagnetic observation data collected by the aircraft and the interference magnetic data generated by the aircraft; S4, obtaining the vector geomagnetic observation data used when the aircraft flies in the target area, as target observation data; S5, determining the interference magnetic data generated by the aircraft based on the linear magnetic observation model for solving the unknown parameters and the target observation data; S6, determining the actual magnetic data of the position of the aircraft when flying in the target area based on the target observation data and the interference magnetic data; S7, navigating the flight of the aircraft in the target area based on the actual magnetic data.

[0028] Based on the existing patent CN117213472A, it can be known that in step S1, the vector geomagnetic observation data collected by the aircraft equipped with the geomagnetic measurement sensor when flying in the calibration area is obtained as the calibration magnetic observation data. The key to this step is to ensure that the flight path of the aircraft in the calibration area is representative, and a specific flight mode (such as five-sided flight) is usually adopted to collect sufficient data. In step S2, the magnetic change rate of the calibration magnetic observation data in the specified direction is determined. Specifically, the magnetic change rate of the vector geomagnetic observation data in a certain direction (such as the cosine direction) is determined. This process helps to identify the impact of the interference magnetic data generated by the aircraft on the original geomagnetic observation data, because the magnetic change rate of the disturbed data in a specific direction will change nonlinearly. In step S3, based on the calibration magnetic observation data and the magnetic change rate, the unknown parameters of the linear magnetic observation model are determined. These parameters reflect the relationship between the geomagnetic observation data collected by the aircraft and the interference magnetic data generated by the aircraft itself. In order to improve the accuracy of the model, it is usually necessary to collect data at multiple time points and fit the model. In step S4, during the navigation phase, the vector geomagnetic observation data collected when the aircraft flies in the target area is obtained as the target observation data. This data will be used in the next step to calculate the actual position of the aircraft in the target area. In step S5, the interference magnetic data generated by the aircraft is determined based on the solved linear magnetic observation model of unknown parameters and the target observation data. This is calculated by substituting the target observation data into the fitted linear magnetic observation model. In step S6, the actual magnetic data of the position of the aircraft when flying in the target area is determined based on the target observation data and the interference magnetic data. This process involves subtracting the interference magnetic data from the target observation data to obtain more accurate geomagnetic data. In step S7, the flight of the aircraft in the target area is navigated based on the actual magnetic data. The navigation system will use the actual magnetic data to determine the current position of the aircraft and generate a navigation path in combination with the destination information.

[0029] Optionally, in an embodiment of the present application, the flight mode of the aircraft in the calibration area (such as five-sided flight) is crucial to the diversity and accuracy of the data. The five-sided flight mode includes actions such as take-off, climb, turn, level flight, descent and landing, which can comprehensively cover data at different altitudes and directions. During the calibration phase, the aircraft needs to collect vector geomagnetic observation data at multiple time points to ensure the accuracy of the model parameters. The data at a single time point may not be sufficient to accurately reflect the interference generated by the aircraft. In order to reduce the impact of noise on model fitting, the collected calibration magnetic observation data and magnetic change rate can be filtered before fitting, such as using techniques such as Kalman filtering.

[0030] In particular, in the above-mentioned geomagnetic navigation method based on vector data, Figure 2This is a schematic flow chart of determining unknown parameters of a linear magnetic observation model based on the calibration magnetic observation data and the magnetic change rate in the geomagnetic navigation method based on vector data of an embodiment of the present application. Figure 2 As shown, based on the calibrated magnetic observation data and the magnetic change rate, the unknown parameters of the linear magnetic observation model are determined, including: S31, determining the magnetic observation matrix according to the magnetic change rate and the component magnetic observation data of the calibrated magnetic observation data in the specified direction; S32, determining the unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data. Accordingly, when determining the unknown parameters of the linear magnetic observation model based on the magnetic observation matrix and the calibrated magnetic observation data, the accuracy of the unknown parameters of the linear magnetic observation model determines the accuracy of the model. Specifically, these parameters reflect the relationship between the geomagnetic observation data collected by the aircraft and the interfering magnetic data generated by the aircraft itself. If these parameters are inaccurate, the model will not be able to correctly separate the interfering magnetic data, thereby affecting the accuracy of the final positioning and navigation.

[0031] Based on this, in the process of determining the unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data, the technical concept of the present application is to analyze the magnetic observation matrix and the calibrated magnetic observation data by using a data processing and analysis algorithm based on artificial intelligence and deep learning to extract the spatial and temporal characteristics of the magnetic observation semantics, and estimate the unknown parameters of the linear magnetic observation model based on the cross-modal joint interactive coding representation between the two. In this way, the unknown parameters of the linear magnetic observation model can be automatically inferred based on the cross-modal joint semantics between the spatial and temporal characteristics of the magnetic observation data, thereby avoiding the inaccuracy of the traditional method of estimating the unknown parameters and fitting the linear magnetic observation model. In this way, the interfering magnetic field generated by the aircraft can be more effectively separated, thereby improving the accuracy and reliability of geomagnetic navigation and positioning.

[0032] Figure 3 A schematic flowchart of determining unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data in the geomagnetic navigation method based on vector data in an embodiment of the present application. Figure 4 This is a schematic diagram of data flow for determining unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data in the geomagnetic navigation method based on vector data in an embodiment of the present application. Figure 3 and Figure 4As shown, specifically, the process of determining the unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data includes: S321, performing feature extraction on the magnetic observation matrix to obtain magnetic observation semantic coding features; S322, performing time series coding on the calibrated magnetic observation data to obtain magnetic observation data time series coding features; S323, performing cross-modal continuous prompt joint coding on the magnetic observation semantic coding features and the magnetic observation data time series coding features to obtain magnetic observation spatiotemporal cross-modal joint coding features; S324, determining the unknown parameters of the linear magnetic observation model based on the magnetic observation spatiotemporal cross-modal joint coding features.

[0033] In the above-mentioned geomagnetic navigation method based on vector data, the S321 performs feature extraction on the magnetic observation matrix to obtain magnetic observation semantic coding features. It should be understood that the original magnetic observation matrix may contain a large amount of redundant information. Feature extraction can reduce the dimension of the data, reduce the computational complexity, and improve processing efficiency. At the same time, through feature extraction, more abstract and advanced spatial features can be extracted from the original magnetic observation matrix. These features can better describe the distribution of the geomagnetic field and help improve the expression ability and accuracy of the model.

[0034] Optionally, in one embodiment of the present application, feature extraction is performed on the magnetic observation matrix to obtain magnetic observation semantic coding features, including: performing hole convolution coding on the magnetic observation matrix to obtain a magnetic observation semantic coding feature map as the magnetic observation semantic coding feature. It should be understood that the magnetic observation matrix is ​​encoded by hole convolution coding to extract the spatial features of magnetic observation semantics. It is worth mentioning that the hole convolution coding method can expand the receptive field without increasing the number of parameters, and capture the spatial information of magnetic observation data in a larger range. Dilated Convolution is a special convolution operation that expands the receptive field by inserting zero values ​​in the convolution kernel, thereby capturing spatial information in a larger range without increasing the number of parameters. Through hole convolution coding, rich spatial features can be extracted from the magnetic observation matrix, and these features can better describe the distribution of the geomagnetic field.

[0035] In the above-mentioned geomagnetic navigation method based on vector data, the S322 performs time-series coding on the calibration magnetic observation data to obtain the time-series coding features of the magnetic observation data. It should be understood that through time-series coding, features in the time dimension can be extracted from the calibration magnetic observation data. These features can capture the dynamic information of the geomagnetic field changing over time, which helps to improve the time perception ability of the model. The changes of the geomagnetic field at different time points may be complex, and time-series coding can help the model better understand these changes, thereby improving the accuracy of the model. During the flight of the aircraft, the geomagnetic observation data will change with time and position. Time-series coding can capture these changes, so that the model can better adapt to the dynamic movement of the aircraft. When the aircraft performs complex maneuvers, the changes in the geomagnetic field may be more drastic. Time-series coding can help the model capture these changes and ensure navigation accuracy in dynamic scenes. Time-series coding can capture the time correlation of the data, reduce the noise impact at a single time point, and make the model more stable and reliable. By extracting time-series features, the model can be better generalized to unseen data, improving the accuracy and reliability of navigation.

[0036] Optionally, in one embodiment of the present application, the calibration magnetic observation data is time-series encoded to obtain the time-series encoding features of the magnetic observation data, including: inputting the calibration magnetic observation data into a sequence encoder based on a bidirectional LSTM model to obtain a magnetic observation data time-series implicit correlation feature vector as the magnetic observation data time-series encoding feature. It should be understood that the bidirectional LSTM model can simultaneously calibrate the forward and backward correlation information of the magnetic observation data in the time dimension, providing a more comprehensive time-series feature. Therefore, through the processing of the sequence encoder based on the bidirectional LSTM model, the forward and backward time-series correlation feature information of the calibration magnetic observation data in the time dimension can be captured, providing a basis for the subsequent estimation of unknown parameters.

[0037] In the above-mentioned geomagnetic navigation method based on vector data, the S323 performs cross-modal continuous prompt joint encoding on the magnetic observation semantic coding features and the magnetic observation data temporal coding features to obtain magnetic observation space-time cross-modal joint coding features. It should be understood that since the magnetic observation semantic coding feature map and the magnetic observation data temporal implicit association feature vector respectively contain the magnetic observation semantic space coding features and magnetic observation data temporal coding features. In geomagnetic navigation, the aircraft needs to determine its position by collecting geomagnetic observation data. However, the interfering magnetic field generated by the aircraft itself will be mixed in the geomagnetic observation data, resulting in positioning errors. In order to accurately separate these interfering magnetic fields, the spatial semantics and temporal semantics of the magnetic observation data play a key role. Therefore, it is necessary to dynamically adjust the behavior of the model according to the different characteristics of the magnetic observation semantics of these two different modalities, so as to enhance the fusion and understanding of cross-modal joint information. However, when processing multimodal data, traditional feature joint coding methods are often difficult to effectively fuse information between different modalities. Based on this, in the technical solution of the present application, the magnetic observation semantic coding features and the magnetic observation data temporal coding features are further cross-modal continuously prompted joint coding to obtain magnetic observation spatiotemporal cross-modal joint coding features. This cross-modal continuous prompt joint coding method enables features between different modalities to be compared and combined in the same representation space to overcome the problem of cross-modal information isolation.

[0038] Optionally, in one embodiment of the present application, the magnetic observation semantic coding features and the magnetic observation data temporal coding features are cross-modally continuously prompted and jointly encoded to obtain magnetic observation spatiotemporal cross-modal joint coding features, including: performing local feature decomposition along the channel dimension of the magnetic observation semantic coding feature map to obtain a set of magnetic observation local feature matrices; performing prompt-perceived cross-modal joint encoding of the set of magnetic observation data temporal implicit association feature vectors and the magnetic observation local feature matrices to obtain the magnetic observation spatiotemporal cross-modal joint coding features.

[0039] Optionally, in one embodiment of the present application, the set of the temporal implicit correlation feature vectors of the magnetic observation data and the set of the magnetic observation local feature matrices are subjected to prompt-aware cross-modal union to obtain the temporal and spatial cross-modal joint coding features of the magnetic observation, including: using the temporal implicit correlation feature vectors of the magnetic observation data as query vectors and using each magnetic observation local feature matrix in the set of the magnetic observation local feature matrix as a key matrix, inputting the query vector and the key matrix into a prompt learning network based on a converter structure to obtain a set of local query prompt semantic coding vectors for magnetic observation spatiotemporal information; using the temporal implicit correlation feature vectors of the magnetic observation data and each magnetic observation local feature matrix in the set of the magnetic observation spatiotemporal information as a key matrix. The query hint semantic encoding vector is input into a cross-modal mask weaving network based on the hint information to obtain a set of magnetic observation spatiotemporal cross-modal local feature mask weight matrices based on the hint information; the positional dot product between each corresponding group of magnetic observation spatiotemporal cross-modal local feature mask weight matrices based on the hint information and the magnetic observation local feature matrix in the set of magnetic observation spatiotemporal cross-modal local feature mask weight matrices based on the hint information and the magnetic observation local feature matrix is ​​calculated to obtain a set of magnetic observation spatiotemporal cross-modal local granularity significant interaction matrices; the set of magnetic observation spatiotemporal cross-modal local granularity significant interaction matrices is feature aggregated to obtain a magnetic observation spatiotemporal cross-modal joint coding feature map as the magnetic observation spatiotemporal cross-modal joint coding feature.

[0040] Optionally, in one embodiment of the present application, the magnetic observation data temporal implicit association feature vector is used as a query vector and each magnetic observation local feature matrix in the set of magnetic observation local feature matrices is used as a key matrix, and the query vector and the key matrix are input into a prompt learning network based on a converter structure to obtain a set of magnetic observation spatiotemporal information local query prompt semantic coding vectors, including: calculating the similarity score between the magnetic observation data temporal implicit association feature vector and each row vector in the magnetic observation local feature matrix to obtain a set of magnetic observation spatiotemporal information similarity scores; normalizing each magnetic observation spatiotemporal information similarity score in the set of magnetic observation spatiotemporal information similarity scores to obtain a set of magnetic observation spatiotemporal information weight coefficients; based on the set of magnetic observation spatiotemporal information weight coefficients, calculating the position-weighted sum of each row vector in the magnetic observation local feature matrix to obtain the magnetic observation spatiotemporal information local query prompt semantic coding vector.

[0041] Optionally, in one embodiment of the present application, each magnetic observation spatiotemporal information similarity score in the set of magnetic observation spatiotemporal information similarity scores is standardized to obtain a set of magnetic observation spatiotemporal information weight coefficients, including: calculating a natural exponential function value with a natural constant e as the base and each magnetic observation spatiotemporal information similarity score in the set of magnetic observation spatiotemporal information similarity scores as the exponential power to obtain a set of magnetic observation spatiotemporal information similarity modulation scores; calculating the set of magnetic observation spatiotemporal information similarity modulation scores point by point to obtain a total sum of the magnetic observation spatiotemporal information similarity scores; dividing the set of magnetic observation spatiotemporal information similarity modulation scores by the total sum of the magnetic observation spatiotemporal information similarity scores by position to obtain the set of magnetic observation spatiotemporal information weight coefficients.

[0042] Optionally, in one embodiment of the present application, each magnetic observation spatiotemporal information local query hint semantic coding vector in the set of the magnetic observation data temporal implicit association feature vector and the magnetic observation spatiotemporal information local query hint semantic coding vector is input into a cross-modal mask weaving network based on hint information to obtain a set of magnetic observation spatiotemporal cross-modal local feature mask weight matrices based on hint information, including: multiplying the transposed vector of the magnetic observation spatiotemporal information local query hint semantic coding vector with the magnetic observation data temporal implicit association feature vector to obtain a magnetic observation spatiotemporal information hint association matrix; dividing the magnetic observation spatiotemporal information hint association matrix by the square root of the scale of the magnetic observation spatiotemporal information hint association matrix by position to obtain a magnetic observation spatiotemporal information hint association modulation matrix; and inputting the magnetic observation spatiotemporal information hint association modulation matrix into a softmax function to obtain the magnetic observation spatiotemporal cross-modal local feature mask weight matrix based on hint information.

[0043] Optionally, in one embodiment of the present application, the similarity scores between the magnetic observation data time series implicit correlation feature vectors and each row vector in the magnetic observation local feature matrix are calculated to obtain a set of magnetic observation spatiotemporal information similarity scores, including: calculating a norm of each row vector in the magnetic observation local feature matrix to obtain a set of magnetic observation local norm values; calculating a norm of the magnetic observation data time series implicit correlation feature vectors to obtain a norm value of the magnetic observation data time series implicit correlation feature vectors; multiplying the norm value of the magnetic observation data time series implicit correlation feature vectors with the set of magnetic observation local norm values ​​by position to obtain a set of magnetic observation spatiotemporal joint norm values; respectively calculating the product between the magnetic observation data time series implicit correlation feature vectors and the transposed vectors of each row vector in the magnetic observation local feature matrix to obtain a set of magnetic observation spatiotemporal correlation values; dividing each group of corresponding magnetic observation spatiotemporal correlation values ​​in the set of magnetic observation spatiotemporal correlation values ​​and the set of magnetic observation spatiotemporal joint norm values ​​by the magnetic observation spatiotemporal joint norm value to obtain the set of magnetic observation spatiotemporal information similarity scores.

[0044] In summary, in the embodiment of the present application, the magnetic observation semantic coding feature and the magnetic observation data temporal coding feature are cross-modal continuously prompted and jointly encoded using the following cross-modal joint coding formula to obtain the magnetic observation spatiotemporal cross-modal joint coding feature;

[0045] Among them, the cross-modal joint encoding formula is:

[0046]

[0047]

[0048]

[0049]

[0050]

[0051] in, is the magnetic observation semantic encoding feature map, , , and are the first, second, and third local feature matrices of the magnetic observation. and A local characteristic matrix of magnetic observations, , , and They are The first, second, and and row vectors, for coding, is the implicit correlation eigenvector of the magnetic observation data time series, yes The transposed vector of is matrix multiplication, is the one-norm of the vector, for and The similarity score of the magnetic observation space-time information between represents the value of the natural exponential function with the natural constant e as the base, yes Middle row vectors, for and The similarity score of the magnetic observation space-time information between for The corresponding magnetic observation spatiotemporal information weight coefficient, yes The number of row vectors in is the semantic encoding vector of local query hints of magnetic observation spatiotemporal information, for The transposed vector of is the scale of the correlation matrix of the spatiotemporal information of magnetic observations, that is, the length of the matrix multiplied by the width of the matrix, yes function, yes The corresponding mask weight matrix of the spatiotemporal cross-modal local feature of magnetic observation based on the prompt information, yes The corresponding mask weight matrix of the spatiotemporal cross-modal local feature of magnetic observation based on the prompt information, For point-by-position multiplication, is the feature concatenation along the channel dimension, It is the spatial-temporal cross-modal joint encoding feature map of the magnetic observation.

[0052] Specifically, first, the local features of the magnetic observation semantic coding feature map are decomposed along the channel dimension to enhance the feature resolution, allowing the algorithm to focus on the cross-modal detail interaction of the magnetic observation spatial semantic features. Then, the temporal implicit correlation feature vector of the magnetic observation data is used as the query vector, and each magnetic observation local feature matrix in the set of magnetic observation local feature matrices is used as the key matrix. The two are processed by a prompt learning network based on a converter structure. The core of the converter structure is the self-attention mechanism, which can automatically learn and dynamically select and highlight the information most relevant to the subsequent unknown parameter estimation and adjust the weight accordingly. The prompt learning network is used to enhance the key semantic interactions between the magnetic observation semantic coding features and the magnetic observation data temporal coding features, as well as the understanding of the relationship between the two different modalities, to ensure that these relationships can be effectively used in subsequent processing.

[0053] Then, each magnetic observation spatiotemporal information local query hint semantic encoding vector in the set of the magnetic observation data temporal implicit association feature vector and the magnetic observation spatiotemporal information local query hint semantic encoding vector is input into a cross-modal mask weaving network based on hint information. This network generates a set of magnetic observation spatiotemporal cross-modal local feature mask weight matrices based on hint information. These mask weight matrices can highlight the important interaction parts between the two modalities of spatial semantics and temporal semantics of magnetic observation data, while suppressing irrelevant information, ensuring that the final output feature map focuses on the most representative area. In other words, the role of the mask weaving network is to create a mask that emphasizes the most relevant part of the modal interaction, reduce the impact of noise, and improve the robustness of the model.

[0054] Subsequently, a point multiplication operation is performed by position based on the set of magnetic observation spatiotemporal cross-modal local feature mask weight matrices based on the prompt information and the corresponding matrices in the set of magnetic observation local feature matrices to obtain a set of magnetic observation spatiotemporal cross-modal local granularity significant interaction matrices. This step further strengthens the degree of fusion of cross-modal information between the magnetic observation semantic coding features and the magnetic observation data temporal coding features, so that the interaction matrix finally obtained can accurately reflect the correlation and importance between the two different modal features of space and time of the magnetic observation data. Finally, by performing feature aggregation on the set of magnetic observation spatiotemporal cross-modal local granularity significant interaction matrices, a magnetic observation spatiotemporal cross-modal joint coding feature map is obtained as the magnetic observation spatiotemporal cross-modal joint coding feature. Through the continuous prompt-assisted learning process, the complex relationship between magnetic observation spatiotemporal data can be better captured, and good performance can still be maintained when facing unseen data.

[0055] In the above-mentioned geomagnetic navigation method based on vector data, the S324 determines the unknown parameters of the linear magnetic observation model based on the spatiotemporal cross-modal joint coding features of the magnetic observation. Optionally, in one embodiment of the present application, the unknown parameters of the linear magnetic observation model are determined based on the spatiotemporal cross-modal joint coding features of the magnetic observation, including: inputting the spatiotemporal cross-modal joint coding feature map of the magnetic observation into a decoder-based parameter estimator to obtain the unknown parameters of the linear magnetic observation model. In other words, the spatiotemporal cross-modal joint coding features of the magnetic observation data are used to perform decoding regression, so as to estimate the unknown parameters of the linear magnetic observation model. In this way, the unknown parameters of the linear magnetic observation model can be automatically inferred based on the cross-modal joint semantics between the spatial features and temporal features of the magnetic observation data, thereby avoiding the inaccuracy of the traditional method of estimating unknown parameters and fitting the linear magnetic observation model. In this way, the interfering magnetic field generated by the aircraft can be more effectively separated, thereby improving the accuracy and reliability of geomagnetic navigation and positioning.

[0056] Optionally, in one embodiment of the present application, the magnetic observation spatiotemporal cross-modal joint coding feature map is input into a decoder-based parameter estimator to obtain unknown parameters of the linear magnetic observation model, including: expanding the magnetic observation spatiotemporal cross-modal joint coding feature map into a magnetic observation spatiotemporal cross-modal joint coding feature vector; using a decoder to decode and regress the magnetic observation spatiotemporal cross-modal joint coding feature vector using the following decoding formula to obtain the unknown parameters of the linear magnetic observation model; wherein the decoding formula is: ,in, is the spatial-temporal cross-modal joint encoding feature vector of the magnetic observation, are the unknown parameters of the linear magnetic observation model, is the weight matrix, Represents matrix multiplication.

[0057] Preferably, inputting the magnetic observation spatiotemporal cross-modal joint encoding feature map into a decoder-based parameter estimator to obtain unknown parameters of the linear magnetic observation model comprises:

[0058] Determine the feature mean and feature variance of the magnetic observation spatiotemporal cross-modal joint coding feature map, and divide the feature mean by the feature variance to obtain a probability statistic value of the magnetic observation spatiotemporal cross-modal joint coding:

[0059]

[0060] in, and They respectively represent the feature mean and feature variance of the spatial-temporal cross-modal joint encoding feature map of the magnetic observation, Represents the probability statistics of the spatiotemporal cross-modal joint encoding of magnetic observations.

[0061] The magnetic observation space-time cross-modal joint coding feature map is point-multiplied with the inverse of the maximum eigenvalue in the magnetic observation space-time cross-modal joint coding feature map to obtain the magnetic observation space-time cross-modal joint coding probability constraint map:

[0062]

[0063] in, represents the spatiotemporal cross-modal joint encoding feature map of the magnetic observation, represents the maximum eigenvalue in the spatiotemporal cross-modal joint encoding feature map of the magnetic observation, It means point multiplication by position. Represents the probability constraint graph of the spatiotemporal cross-modal joint encoding of magnetic observations.

[0064] After dot-wise addition of the magnetic observation space-time cross-modal joint coding probability constraint graph and the magnetic observation space-time cross-modal joint coding probability statistical value, the logarithmic value with base 2 is calculated to obtain the magnetic observation space-time cross-modal joint coding information interaction graph:

[0065]

[0066] in, represents the probability constraint graph of the spatiotemporal cross-modal joint encoding of magnetic observations, represents the probability statistics of the spatial-temporal cross-modal joint coding of magnetic observations, Indicates adding by position point. Represents the information interaction graph of the cross-modal joint encoding of magnetic observations in space and time.

[0067] After performing point subtraction on the magnetic observation space-time cross-modal joint coding probability constraint graph with the magnetic observation space-time cross-modal joint coding probability statistics, the inverse of each eigenvalue is calculated to obtain the magnetic observation space-time cross-modal joint coding sequence constraint graph:

[0068]

[0069] in, represents the probability constraint graph of the spatiotemporal cross-modal joint encoding of magnetic observations, represents the probability statistics of the spatial-temporal cross-modal joint coding of magnetic observations, It means to subtract by position point. Represents the sequential constraint graph of the spatiotemporal cross-modal joint encoding of magnetic observations.

[0070] The magnetic observation space-time cross-modal joint coding information interaction graph and the magnetic observation space-time cross-modal joint coding sequence constraint graph are interpolated to obtain an optimized magnetic observation space-time cross-modal joint coding feature graph;

[0071] The optimized magnetic observation spatiotemporal cross-modal joint encoding feature map is input into a decoder-based parameter estimator to obtain the unknown parameters of the linear magnetic observation model.

[0072] That is, when the magnetic observation semantic coding feature map and the magnetic observation data temporal implicit association feature vector respectively represent the semantic association features of the spatial dimension of magnetic observation data and the semantic association features of the temporal dimension of magnetic observation data, when performing cross-modal joint operation assisted by continuous prompts, the magnetic observation spatiotemporal cross-modal joint coding feature map will also have the cross-modal joint distribution diversity assisted by continuous prompts under the heterogeneous distribution of the temporal dimension and spatial dimension of magnetic observation data. As a result, when the magnetic observation spatiotemporal cross-modal joint coding feature map is decoded by a decoder, it will affect the accuracy of the decoding result.

[0073] Therefore, considering that when the weight matrix of the decoder acts on the magnetic observation space-time cross-modal joint coding feature vector after the magnetic observation space-time cross-modal joint coding feature map is expanded, the distribution diversity of the magnetic observation space-time cross-modal joint coding feature map causes the uncertainty of the weight stimulation parameters of the weight matrix of the decoder, which makes the posterior probability density inference of the magnetic observation space-time cross-modal joint coding feature map through the action of the weight matrix missing, which will affect the accuracy of the decoding result.

[0074] In this case, the probability statistical characteristics of the magnetic observation spatiotemporal cross-modal joint encoding feature map are used to simulate the mesoscale interaction structure under probability constraints between the eigenvalue scale and the feature map scale of the magnetic observation spatiotemporal cross-modal joint encoding feature map, so as to construct a bidirectional latent variable motif based on the mesoscale short sequence relative to the probability statistical value to perform mesoscale bidirectional migration, and perform a posteriori recovery based on the sequence constraints on the interactive information, thereby improving the convergence effect in the probability density domain and improving the accuracy of the unknown parameters of the linear magnetic observation model obtained by the decoder-based parameter estimator of the magnetic observation spatiotemporal cross-modal joint encoding feature map.

[0075] In summary, the geomagnetic navigation method based on vector data according to the embodiment of the present application is explained, which analyzes the magnetic observation matrix and calibrated magnetic observation data by adopting a data processing and analysis algorithm based on artificial intelligence and deep learning to extract the spatial and temporal characteristics of the magnetic observation semantics, and estimates the unknown parameters of the linear magnetic observation model based on the cross-modal joint interactive coding representation between the two. In this way, the unknown parameters of the linear magnetic observation model can be automatically inferred based on the cross-modal joint semantics between the spatial characteristics and temporal characteristics of the magnetic observation data, thereby avoiding the inaccuracy of the traditional method of estimating unknown parameters and fitting the linear magnetic observation model. In this way, the interfering magnetic field generated by the aircraft can be more effectively separated, thereby improving the accuracy and reliability of geomagnetic navigation and positioning.

[0076] Figure 5 FIG. 1 is a schematic block diagram of a geomagnetic navigation system based on vector data according to an embodiment of the present application. Figure 5 As shown, the geomagnetic navigation system 100 based on vector data includes: a magnetic observation matrix feature extraction module 110, used to extract features from the magnetic observation matrix to obtain magnetic observation semantic coding features; a calibrated magnetic observation data time series coding module 120, used to perform time series coding on the calibrated magnetic observation data to obtain magnetic observation data time series coding features; a cross-modal continuous prompt joint coding module 130, used to perform cross-modal continuous prompt joint coding on the magnetic observation semantic coding features and the magnetic observation data time series coding features to obtain magnetic observation space-time cross-modal joint coding features; an unknown parameter determination module 140, used to determine the unknown parameters of the linear magnetic observation model based on the magnetic observation space-time cross-modal joint coding features.

[0077] Here, those skilled in the art will appreciate that the specific operations of the various modules and units in the above-mentioned geomagnetic navigation system based on vector data have been described in detail above. Figures 1 to 4 The description of the geomagnetic navigation method based on vector data has been introduced in detail, and therefore, its repeated description will be omitted.

[0078] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0079] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules 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 may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0080] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0081] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be considered exemplary and non-restrictive in all respects, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all variations that fall within the meaning and scope of the equivalent elements of the claims be included in the present invention.

[0082] In addition, it is obvious that the word "comprising" does not exclude other elements or steps and the singular does not exclude the plural.

[0083] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A geomagnetic navigation method based on vector data, comprising: Acquire vector geomagnetic observation data collected when an aircraft equipped with a geomagnetic measurement sensor is flying as calibration magnetic observation data; determine the magnetic change rate of the calibration magnetic observation data in a specified direction; determine unknown parameters of a linear magnetic observation model based on the calibration magnetic observation data and the magnetic change rate, wherein the parameters are used to represent the relationship between the geomagnetic observation data collected by the aircraft and the interference magnetic data generated by the aircraft; acquire vector geomagnetic observation data used when the aircraft flies in a target area as target observation data; determine the interference magnetic data generated by the aircraft based on the linear magnetic observation model for solving the unknown parameters and the target observation data; determine the actual magnetic data of the position of the aircraft when flying in the target area based on the target observation data and the interference magnetic data; and navigate the flight of the aircraft in the target area based on the actual magnetic data; Wherein, based on the calibrated magnetic observation data and the magnetic change rate, determining the unknown parameters of the linear magnetic observation model includes: determining the magnetic observation matrix according to the magnetic change rate and the component magnetic observation data of the calibrated magnetic observation data in the specified direction; determining the unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data; It is characterized in that determining the unknown parameters of the linear magnetic observation model according to the magnetic observation matrix and the calibrated magnetic observation data includes: Performing feature extraction on the magnetic observation matrix to obtain magnetic observation semantic coding features; Performing time series encoding on the calibration magnetic observation data to obtain time series encoding features of the magnetic observation data; Performing cross-modal continuous prompt joint encoding on the magnetic observation semantic encoding features and the magnetic observation data temporal encoding features to obtain magnetic observation spatiotemporal cross-modal joint encoding features; Determining unknown parameters of the linear magnetic observation model based on the spatiotemporal cross-modal joint encoding features of the magnetic observation; Performing feature extraction on the magnetic observation matrix to obtain magnetic observation semantic coding features, including: performing hole convolution coding on the magnetic observation matrix to obtain a magnetic observation semantic coding feature map as the magnetic observation semantic coding feature; Performing time series encoding on the calibrated magnetic observation data to obtain time series encoding features of the magnetic observation data, including: inputting the calibrated magnetic observation data into a sequence encoder based on a bidirectional LSTM model to obtain a time series implicit correlation feature vector of the magnetic observation data as the time series encoding feature of the magnetic observation data; The magnetic observation semantic coding feature and the magnetic observation data temporal coding feature are jointly coded in a cross-modal continuous prompt manner to obtain a magnetic observation spatiotemporal cross-modal joint coding feature, including: Performing local feature decomposition along the channel dimension on the magnetic observation semantic coding feature map to obtain a set of magnetic observation local feature matrices; The set of the temporal implicit correlation feature vector of the magnetic observation data and the local feature matrix of the magnetic observation is subjected to prompt-perceived cross-modal union to obtain the spatiotemporal cross-modal joint coding feature of the magnetic observation.

2. The geomagnetic navigation method based on vector data according to claim 1, characterized in that: The set of the temporal implicit correlation feature vector of the magnetic observation data and the local feature matrix of the magnetic observation is combined with the prompt perception cross-modality to obtain the spatiotemporal cross-modality joint coding feature of the magnetic observation, including: Taking the temporal implicit correlation feature vector of the magnetic observation data as a query vector and taking each magnetic observation local feature matrix in the set of the magnetic observation local feature matrix as a key matrix, the query vector and the key matrix are input into a prompt learning network based on a converter structure to obtain a set of magnetic observation spatiotemporal information local query prompt semantic encoding vectors; Inputting each magnetic observation spatiotemporal information local query prompt semantic coding vector in the set of the magnetic observation data temporal implicit association feature vector and the magnetic observation spatiotemporal information local query prompt semantic coding vector into a cross-modal mask weaving network based on prompt information to obtain a set of magnetic observation spatiotemporal cross-modal local feature mask weight matrices based on prompt information; Calculate the set of magnetic observation spatiotemporal cross-modal local feature mask weight matrices based on the prompt information and the set of magnetic observation local feature matrices, and multiply each corresponding magnetic observation spatiotemporal cross-modal local feature mask weight matrix based on the prompt information and the magnetic observation local feature matrix by the position point to obtain the set of magnetic observation spatiotemporal cross-modal local granularity significant interaction matrices; The set of the magnetic observation space-time cross-modal local granularity significant interaction matrices is feature aggregated to obtain a magnetic observation space-time cross-modal joint coding feature map as the magnetic observation space-time cross-modal joint coding feature.

3. The geomagnetic navigation method based on vector data according to claim 2, characterized in that: Taking the temporal implicit correlation feature vector of the magnetic observation data as a query vector and each magnetic observation local feature matrix in the set of the magnetic observation local feature matrix as a key matrix, the query vector and the key matrix are input into a prompt learning network based on a converter structure to obtain a set of magnetic observation spatiotemporal information local query prompt semantic encoding vectors, including: Calculating the similarity scores between the temporal implicit correlation feature vector of the magnetic observation data and each row vector in the magnetic observation local feature matrix to obtain a set of similarity scores of the spatiotemporal information of the magnetic observation; Standardizing each magnetic observation spatiotemporal information similarity score in the set of magnetic observation spatiotemporal information similarity scores to obtain a set of magnetic observation spatiotemporal information weight coefficients; Based on the set of weight coefficients of the magnetic observation spatiotemporal information, the position-weighted sum of each row vector in the magnetic observation local feature matrix is ​​calculated to obtain the local query prompt semantic coding vector of the magnetic observation spatiotemporal information.

4. The geomagnetic navigation method based on vector data according to claim 3, characterized in that: The respective magnetic observation spatiotemporal information similarity scores in the set of magnetic observation spatiotemporal information similarity scores are normalized to obtain a set of magnetic observation spatiotemporal information weight coefficients, including: Calculate a natural exponential function value with a natural constant e as the base and each magnetic observation space-time information similarity score in the set of magnetic observation space-time information similarity scores as an exponential power to obtain a set of magnetic observation space-time information similarity modulation scores; Calculating the set of magnetic observation spatiotemporal information similarity modulation scores position by position to obtain a total sum of magnetic observation spatiotemporal information similarity scores; The set of magnetic observation spatiotemporal information similarity modulation scores is divided by the total sum of the magnetic observation spatiotemporal information similarity scores by position to obtain the set of magnetic observation spatiotemporal information weight coefficients.

5. The geomagnetic navigation method based on vector data according to claim 4, characterized in that: Inputting each magnetic observation spatiotemporal information local query hint semantic coding vector in the set of the magnetic observation data temporal implicit association feature vector and the magnetic observation spatiotemporal information local query hint semantic coding vector into a cross-modal mask weaving network based on hint information to obtain a set of magnetic observation spatiotemporal cross-modal local feature mask weight matrices based on hint information, including: Multiplying the transposed vector of the magnetic observation spatiotemporal information local query prompt semantic coding vector by the magnetic observation data temporal implicit association feature vector to obtain a magnetic observation spatiotemporal information prompt association matrix; Dividing the magnetic observation space-time information prompt association matrix by the square root of the scale of the magnetic observation space-time information prompt association matrix by position to obtain a magnetic observation space-time information prompt association modulation matrix; The magnetic observation spatiotemporal information prompt association modulation matrix is ​​input into the softmax function to obtain the magnetic observation spatiotemporal cross-modal local feature mask weight matrix based on the prompt information.

6. The geomagnetic navigation method based on vector data according to claim 5, characterized in that: Based on the magnetic observation space-time cross-modal joint coding characteristics, the unknown parameters of the linear magnetic observation model are determined, including: inputting the magnetic observation space-time cross-modal joint coding feature map into a decoder-based parameter estimator to obtain the unknown parameters of the linear magnetic observation model.

7. A geomagnetic navigation system based on vector data according to the geomagnetic navigation method based on vector data according to any one of claims 1 to 6, characterized in that: A magnetic observation matrix feature extraction module, used for extracting features from the magnetic observation matrix to obtain magnetic observation semantic coding features; A calibration magnetic observation data time series encoding module, used for performing time series encoding on the calibration magnetic observation data to obtain a time series encoding feature of the magnetic observation data; A cross-modal continuous prompt joint encoding module, used for performing cross-modal continuous prompt joint encoding on the magnetic observation semantic encoding feature and the magnetic observation data temporal encoding feature to obtain a magnetic observation spatiotemporal cross-modal joint encoding feature; The unknown parameter determination module is used to determine the unknown parameters of the linear magnetic observation model based on the spatiotemporal cross-modal joint encoding characteristics of the magnetic observation.

Images