A satellite orbit kilometer index parameter prediction method, device, equipment and medium

Abstract

The application discloses a satellite orbit kilometer marker parameter prediction method and device, equipment and medium, and relates to the technical field of satellite data processing. The method comprises the following steps: converting longitude and latitude in satellite positioning data into UTM coordinates; performing vector projection on the UTM coordinates on a line segment to determine the kilometer markers of each key position point; performing multi-time-step lag feature splicing and difference processing on the satellite positioning data and the kilometer markers respectively to generate the final feature matrix of the satellite positioning data and the kilometer markers at each key position point, and performing robust standardization; training a double-layer long short-term memory network to obtain a kilometer marker prediction model; inputting the satellite feature sequence of a to-be-measured position point on a railway track section into the kilometer marker prediction model to output the kilometer marker of the to-be-measured position point on the railway track section. The application can effectively improve the accuracy of kilometer marker parameter prediction.

Description

Technical Field

[0001] This application relates to the field of satellite data processing technology, and in particular to a method, apparatus, equipment and medium for predicting satellite orbit kilometer marker parameters. Background Technology

[0002] With the rapid advancement of satellite navigation and space situational awareness technologies, high-precision satellite orbit prediction has become a core technology for ensuring the on-orbit safety of spacecraft and improving the effectiveness of space missions.

[0003] Traditional satellite orbit parameter prediction methods (such as autoregressive integral moving average and linear regression) struggle to capture nonlinear time dependencies. Satellite data contains high-frequency noise and outliers, resulting in poor robustness of prediction models. Single-time-step features cannot fully characterize the dynamics of orbital changes, and conventional data standardization methods (such as Z-score) are sensitive to data distribution and easily affected by extreme values. Therefore, there is an urgent need to overcome the accuracy bottleneck in existing satellite orbit parameter prediction methods. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, equipment, and medium for predicting satellite orbit kilometer marker parameters, which can effectively improve the accuracy of kilometer marker parameter prediction.

[0005] To achieve the above objectives, this application provides the following solution.

[0006] Firstly, this application provides a method for predicting satellite orbit kilometer marker parameters, comprising: acquiring satellite positioning data of multiple key locations on a railway track section obtained by satellite positioning, and converting the latitude and longitude in the data into UTM coordinates; based on the line segments between known kilometer marker locations on the railway track section, vector-projecting the UTM coordinates onto the line segments to determine the kilometer markers at each key location point; concatenating the satellite positioning data and kilometer markers at each key location point with lag features obtained over multiple time steps to obtain extended feature matrices for the satellite positioning data and kilometer markers at each key location point; and determining the differential features and moving average features of each extended feature matrix. The system first identifies the features of each extended feature matrix and generates the final feature matrix from its differential and moving average features. Robust standardization is then applied to each final feature matrix to obtain the standard feature matrices of satellite positioning data and kilometer markers at each key location point. Using the satellite feature sequences at each time step in the standard feature matrices of satellite positioning data at each key location point as input, and the kilometer markers at each time step in the standard feature matrices of kilometer markers at each key location point as labels, a two-layer long short-term memory network is trained to obtain the kilometer marker prediction model. Finally, the satellite feature sequences of the locations to be measured on the railway track section are input into the kilometer marker prediction model, and the kilometer markers of the locations to be measured on the railway track section are output.

[0007] Optionally, the latitude and longitude in the data can be converted to UTM coordinates, specifically including:

[0008] Use longitude from satellite positioning data to determine the UTM partition number;

[0009] Based on the ellipsoid parameters and the latitude from satellite positioning data, using the formula... and Determine the radius of curvature; where, Let be the radius of curvature. For the semi-major axis of the ellipsoid, For eccentricity, Latitude For ellipsoidal parameters;

[0010] Based on the radius of curvature, using the formula Determine the east coordinates of the UTM; where, For UTM East coordinates, For eastward pseudo-offset, As a scaling factor, For longitude difference, , Longitude The longitude of the central meridian. The square term of the latitude tangent. , For correction terms related to eccentricity and latitude, ;

[0011] Based on the radius of curvature, using the formula Determine the north coordinates of the UTM; where, UTM North coordinates The distance on an ellipsoid from the equator to latitude is... ;

[0012] The UTM partition number, UTM east coordinate, and UTM north coordinate together constitute the UTM coordinate.

[0013] Optionally, based on the line segments between known kilometer markers on the railway track section, the UTM coordinates are vector-projected onto the line segments to determine the kilometer markers for each key location point, specifically including:

[0014] Determine the line segments between all known kilometer markers on the railway track section;

[0015] Based on the UTM coordinates of each key location point and the endpoints of each line segment, the formula is used. Determine the projection parameters of the UTM coordinates of each key location point onto each line segment; where, For projection parameters, , The distance from the UTM coordinates of the critical location point to one endpoint of the line segment. , UTM coordinates of key location points Let the coordinates of one endpoint of the line segment be . The distance from the UTM coordinates of the critical location point to the other endpoint of the line segment. , Let the coordinates be the coordinates of the other endpoint of the line segment;

[0016] Based on the projection parameters, using the formula Determine the projection coordinates of the UTM coordinates of each key location point onto each line segment; where, The coordinates of the projection point;

[0017] Based on the UTM coordinates and projection point coordinates of each key location point, select the projection point coordinates and line segment corresponding to the minimum projection distance, and use the selected line segment as the projection line segment.

[0018] If the coordinates of the selected projection point lie on the projection line segment, then use the formula Determine the kilometer markers for each key location; where, For key location kilometer markers, The kilometer marker at one endpoint of the projected line segment. The kilometer marker at the other endpoint of the projected line segment;

[0019] If the selected projection point coordinates are outside the projection line segment, then the kilometer marker of the endpoint on the projection line segment closest to the projection point coordinates will be determined as the kilometer marker of each key location point.

[0020] Optionally, the satellite positioning data of each key location point are concatenated with lag features from multiple time steps to obtain an extended feature matrix of the satellite positioning data at each key location point, specifically including:

[0021] The satellite positioning data with multiple time steps of lag are used to construct a lag feature vector;

[0022] By concatenating the satellite positioning data of each key location point with the lag feature vector, an extended feature matrix of the satellite positioning data at each key location point is obtained.

[0023] Optionally, the difference features and moving average features of each extended feature matrix are determined, and the final feature matrix is ​​generated from each extended feature matrix and its difference features and moving average features, specifically including:

[0024] Using formula Calculate the difference features of each extended feature matrix; where, For time step Difference features of the extended feature matrix, For time step The extended feature matrix, For time step The extended feature matrix;

[0025] Using formula Calculate the moving average characteristic of each extended characteristic matrix; where, For time step Moving average feature of extended feature matrix, For time step The extended feature matrix, The offset within the window. =0,1,2;

[0026] The final feature matrix is ​​obtained by merging the extended feature matrices with the corresponding difference features and moving average features.

[0027] Optionally, robust normalization is performed on each final feature matrix to obtain standard feature matrices for both satellite positioning data and kilometer markers at each key location point, specifically including:

[0028] Determine the median and interquartile range of each final feature matrix;

[0029] Based on the median and the interquartile range, using the formula The final feature matrices are standardized to obtain the standard feature matrices of satellite positioning data and kilometer markers at each key location point; where, For the final characteristic matrix, the th The standardized feature values ​​of each feature For the th in the final feature matrix One characteristic, For the final characteristic matrix, the th The median of the features, For the final characteristic matrix, the th Interquartile range of each feature.

[0030] Optionally, the two-layer long short-term memory network includes: a first long short-term memory network, a first dropout layer, a second long short-term memory network, a second dropout layer, and a fully connected layer connected in sequence;

[0031] The dual-layer long short-term memory network has an adaptive learning rate.

[0032] Secondly, this application provides a satellite orbit kilometer marker parameter prediction device, including: a conversion module, a vector projection module, a feature stitching module, a feature merging module, a standard module, a training module, and an application module.

[0033] The system comprises the following modules: a conversion module, which acquires satellite positioning data of multiple key locations on the railway track section obtained by satellite positioning, and converts the latitude and longitude of the data into UTM coordinates; a vector projection module, which projects the UTM coordinates onto the line segments between known kilometer markers on the railway track section to determine the kilometer markers of each key location; a feature stitching module, which stitches multi-time-step lag features onto the satellite positioning data and kilometer markers of each key location to obtain the extended feature matrices of the satellite positioning data and kilometer markers at each key location; and a feature merging module, which determines the difference features and moving average features of each extended feature matrix and merges the extended features. The system generates the final feature matrix using matrices, their difference features, and moving average features. A standard module robustly standardizes each final feature matrix to obtain standard feature matrices for both satellite positioning data and kilometer markers at key locations. A training module uses the satellite feature sequences at each time step of the standard feature matrices for satellite positioning data at key locations as input, and the kilometer markers at each time step of the standard feature matrices for kilometer markers at key locations as labels, to train a two-layer long short-term memory network and obtain a kilometer marker prediction model. An application module inputs the satellite feature sequences of the locations to be measured on the railway track section into the kilometer marker prediction model and outputs the kilometer markers for the locations to be measured on the railway track section.

[0034] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the satellite orbit kilometer parameter prediction method described in any one of the above-described methods.

[0035] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the satellite orbit kilometer parameter prediction method described above.

[0036] According to the specific embodiments provided in this application, this application has the following technical effects.

[0037] This application provides a method, apparatus, device, and medium for predicting satellite orbit kilometer marker parameters. It utilizes a dual-layer long short-term memory (LSTM) network to predict kilometer markers. The dual-layer LSTM network enables hierarchical temporal feature extraction, effectively capturing nonlinear temporal dependencies. During training, the dual-layer LSTM network can indirectly learn internal representations insensitive to high-frequency noise and outliers, thus improving prediction robustness. By performing multi-time-step lag feature concatenation on satellite positioning data and kilometer markers at each key location point, it overcomes the limitation of single-time-step features in fully representing the dynamic characteristics of orbital changes. Robust normalization is applied to the final feature matrices of both satellite positioning data and kilometer markers at each key location point, addressing the problem of conventional data normalization methods being sensitive to data distribution and susceptible to extreme values. Combining these technical means, this application effectively improves the accuracy of kilometer marker parameter prediction. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a flowchart illustrating a method for predicting satellite orbit kilometer marker parameters provided in an embodiment of this application.

[0040] Figure 2 This is a schematic diagram of the kilometer marker implementation module for determining key location points provided in an embodiment of this application.

[0041] Figure 3 A schematic diagram of UTM conversion and kilometer marker projection provided for an embodiment of this application.

[0042] Figure 4 This is a schematic diagram of the implementation module for obtaining the kilometer marker prediction model provided in an embodiment of this application.

[0043] Figure 5 This is a schematic diagram comparing the predicted and actual values ​​of the kilometer marker prediction model provided in the embodiments of this application.

[0044] Figure 6 This is a schematic diagram provided for an embodiment of this application.

[0045] Figure 7 This diagram illustrates the correlation between the predicted and actual values ​​of the kilometer marker prediction model provided in this application embodiment.

[0046] Figure 8 This is a schematic diagram of the functional modules of a satellite orbit kilometer marker parameter prediction device provided in an embodiment of this application.

[0047] Figure 9 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0048] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0049] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0050] In one exemplary embodiment, such as Figure 1 As shown, a method for predicting satellite orbit kilometer marker parameters is provided, including the following steps 101 to 107.

[0051] Step 101: Obtain satellite positioning data of multiple key locations on the railway track section measured by satellite positioning, and convert the latitude and longitude in the data into UTM coordinates.

[0052] Step 102: Based on the line segments between the known kilometer markers on the railway track section, project the UTM coordinates onto the line segments to determine the kilometer markers of each key location point.

[0053] Step 103: Concatenate the lag features of the satellite positioning data and kilometer markers at each key location point with the features of multiple time steps to obtain the extended feature matrices of the satellite positioning data and kilometer markers at each key location point.

[0054] Step 104: Determine the difference features and moving average features of each extended feature matrix, and generate the final feature matrix from each extended feature matrix and its difference features and moving average features.

[0055] Step 105: Perform robust normalization on each final feature matrix to obtain the standard feature matrices of satellite positioning data and kilometer markers at each key location point.

[0056] Step 106: Using the satellite feature sequence of each time step in the standard feature matrix of satellite positioning data at each key location point as input, and the kilometer markers of each time step in the standard feature matrix of kilometer markers at each key location point as labels, train a two-layer long short-term memory network to obtain the kilometer marker prediction model.

[0057] Step 107: Input the satellite feature sequence of the location point to be measured on the railway track section into the kilometer marker prediction model, and output the kilometer marker of the location point to be measured on the railway track section.

[0058] By implementing steps 101 to 107 above, satellite orbit kilometer values ​​were predicted, and the accuracy of the prediction parameters was effectively improved in the process of predicting orbit kilometer values.

[0059] In another exemplary embodiment of this application, the satellite positioning data is first preprocessed: the satellite positioning data is arranged in order of latitude and longitude coordinates to achieve temporal sequence. Then, using the WGS84 ellipsoid parameters, the UTM (Universal Transverse Mercator) algorithm is called to convert the latitude and longitude into UTM coordinates. The format of UTM coordinates is: area code + east coordinate + north coordinate. The conversion of latitude and longitude in the satellite positioning data to UTM coordinates in step 101 above can be replaced by steps 201 to 205.

[0060] Step 201: Determine the UTM partition number using the longitude from the satellite positioning data.

[0061] Step 202: Based on the ellipsoid parameters and the latitude from the satellite positioning data, use the formula... and Determine the radius of curvature; where, Let be the radius of curvature. For the semi-major axis of the ellipsoid, For eccentricity, Latitude (radians) These are the parameters of the ellipsoid. Used in UTM conversion to correct curvature in the latitudinal direction. and All are ellipsoidal parameters.

[0062] Step 203: Based on the radius of curvature, use the formula Determine the east coordinates of the UTM; where, For UTM East coordinates, For eastward pseudo-offset, As a scaling factor, For longitude difference, , Longitude (in radians) The longitude of the central meridian. The square term of the latitude tangent. , For correction terms related to eccentricity and latitude, . Determined by the UTM partition number.

[0063] Step 204: Based on the radius of curvature, use the formula Determine the north coordinates of the UTM; where, UTM North coordinates The distance on an ellipsoid from the equator to latitude is... . In UTM north coordinate calculations, it is used to determine the north-south position of a point relative to the equator.

[0064] UTM North Coordinates The formula expands with an even number of terms, and all expansions are calculated using WGS84 ellipsoidal parameters. The ellipsoidal distance from the equator to latitude... The formula for calculating the arc length of a meridian is the standard meridian arc length formula, which is usually written in the form of a Fourier series.

[0065] Step 205: Combine the UTM partition number, UTM east coordinate, and UTM north coordinate to form the UTM coordinate.

[0066] In another exemplary embodiment of this application, the nearest point is determined by vector projection, and the projection parameters are calculated using vector dot product, limiting the range to the line segment. Based on linear interpolation of the kilometer markers at the start and end points of the line segment, the nearest projection point or the kilometer marker corresponding to the endpoint is selected from all line segments. Therefore, step 102 can be replaced by steps 301 to 306.

[0067] Step 301: Determine the line segments between all known kilometer markers on the railway track section.

[0068] Step 302: Based on the UTM coordinates of each key location point and the endpoints of each line segment, use the formula... Determine the projection parameters of the UTM coordinates of each key location point onto each line segment; where, For projection parameters, , The distance from the UTM coordinates of the critical location point to one endpoint of the line segment. , UTM coordinates of key location points Let the coordinates of one endpoint of the line segment be . The distance from the UTM coordinates of the critical location point to the other endpoint of the line segment. , Let be the coordinates of the other endpoint of the line segment.

[0069] Step 303: Based on the projection parameters, use the formula Determine the projection coordinates of the UTM coordinates of each key location point onto each line segment; where, These are the coordinates of the projection point.

[0070] Step 304: Based on the UTM coordinates and projection point coordinates of each key location point, select the projection point coordinates and line segment corresponding to the minimum projection distance, and use the selected line segment as the projection line segment.

[0071] Step 305: If the coordinates of the selected projection point lie on the projection line segment, then use the formula... Determine the kilometer markers for each key location; where, For key location kilometer markers, The kilometer marker at one endpoint of the projected line segment. The kilometer marker is the other endpoint of the projected line segment.

[0072] Step 306: If the selected projection point coordinates are outside the projection line segment, then the kilometer marker of the endpoint on the projection line segment closest to the projection point coordinates is determined as the kilometer marker of each key location point.

[0073] The kilometer markers at each key location can be represented as: In the formula, These are kilometer markers for key locations.

[0074] For example, such as Figure 2 As shown, steps 101 to 102 are implemented by an input module, a UTM conversion module, a cumulative distance calculation module, a projection calculation module, and an output module. The input module receives key points of the route (longitude, latitude, and kilometer markers) and the set of points to be converted; the UTM conversion module calls the UTM algorithm to convert latitude and longitude into planar coordinates; the cumulative distance calculation module generates the mileage information of the path by accumulating line segments; the projection calculation module projects the points to be measured onto the nearest line segment and calculates the kilometer markers within the effective interval; the output module outputs the kilometer marker results in a formatted manner (e.g., K071+500).

[0075] Figure 3 The UTM conversion and kilometer marker projection provided in this application embodiment visualizes the results of converting satellite latitude and longitude coordinates into orbital kilometer markers. Figure 3 Part (a) shows the spatial projection relationship of geographic coordinates. Figure 3 Part (b) shows the UTM coordinate space projection relationship, in Figure 3 In part (b), the projection points cover the points to be measured.

[0076] In another exemplary embodiment of this application, if all satellite positioning data and kilometer markers are set with lag features, then step 103 above, which splices the satellite positioning data of each key location point with lag features of multiple time steps to obtain the extended feature matrix of satellite positioning data at each key location point, can be replaced by steps 401 to 402.

[0077] Step 401: Construct a lag feature vector from the satellite positioning data with multiple time steps of lag.

[0078] For example, the matrix corresponding to the lagged eigenvector It can be represented as:

[0079] ;

[0080] In the formula, This represents the total number of samples. This represents the maximum number of lag steps. For at any time The satellite positioning data, as in this application, consists of three feature values ​​(longitude, latitude, and altitude), so each It contains 3 values. .

[0081] Step 402: Concatenate the satellite positioning data of each key location point with the lag feature vector to obtain the extended feature matrix of the satellite positioning data at each key location point.

[0082] For example, if there is a lag of 3 time steps, the lag feature vector is:

[0083] ;

[0084] in, For the first 3 time steps A matrix composed of satellite positioning data (raw features), with dimensions of 3×3; , , These are the original feature vector values ​​for the first three time steps.

[0085] The extended feature matrix of satellite positioning data at each key location point is represented as follows:

[0086] ;

[0087] in, The merged extended feature vector has a dimension of . (Original 3D + Lag 9D). These are the original feature vector values, with dimension 1. . This is a matrix flattening operation that converts a 3×3 matrix into a 9×1 vector.

[0088] The steps for performing multi-time-step lag feature stitching on the kilometer markers to obtain the extended feature matrix of the kilometer markers at each key location point are the same as steps 401 to 402.

[0089] In another exemplary embodiment of this application, short-term trends are captured by difference operations within a time window, and moving average fusion is performed to smooth out high-frequency noise using a sliding window mean. Therefore, the generation of the final feature matrix in step 104 can be replaced by steps 501-503.

[0090] Step 501: Using the formula Calculate the difference features of each extended feature matrix; where, For time step Difference features of the extended feature matrix, For time step The extended feature matrix, For time step The extended feature matrix.

[0091] Step 502: Using the formula Calculate the moving average characteristic of each extended characteristic matrix; where, For time step Moving average feature of extended feature matrix, For time step The extended feature matrix, The offset within the window. =0,1,2.

[0092] Step 503: Merge each extended feature matrix with the corresponding difference feature and moving average feature to obtain the final feature matrix.

[0093] Difference features refer to the difference between the current value and the historical value, and are used to capture short-term trends. The window size is 3, used to smooth noise and reflect long-term trends.

[0094] For example, based on the above lag of 3 time steps... The final feature matrix can be expressed as:

[0095] ;

[0096] in, The final feature vector after merging has a dimension of . (Extended features 12-dimensional + difference 12-dimensional + moving average 12-dimensional).

[0097] In another exemplary embodiment of this application, the above step 105, which robustly normalizes each final feature matrix to obtain the standard feature matrix of each satellite positioning data and kilometer marker at each key location point, can be replaced by the following steps 601 to 602.

[0098] Step 601: Determine the median and interquartile range of each final feature matrix.

[0099] Step 602: Based on the median and the interquartile range, use the formula... The final feature matrices are standardized to obtain the standard feature matrices of satellite positioning data and kilometer markers at each key location point; where, For the final characteristic matrix, the th The standardized feature values ​​of each feature For the th in the final feature matrix One characteristic, For the final characteristic matrix, the th The median of each feature. For the final characteristic matrix, the th Interquartile range (IQR) of each feature. , , For the final characteristic matrix, the th Data for each feature.

[0100] In another exemplary embodiment of this application, the two-layer long short-term memory network includes: a first long short-term memory network, a first dropout layer, a second long short-term memory network, a second dropout layer, and a fully connected layer connected sequentially. The first long short-term memory network outputs the complete sequence, and then connects to the first dropout layer to prevent overfitting, with a ratio of 0.4; the second long short-term memory network outputs the last time step, and then connects to the second dropout layer, with a ratio of 0.3; finally, the fully connected layer progressively reduces the dimensionality to 1 dimension for the output.

[0101] The improvements of the two-layer long short-term memory network compared to the LSTM (Long Short-Term Memory) network include: the use of a two-layer cascaded LSTM structure, dynamic regularization, adaptive learning rate scheduling, and prediction error compensation.

[0102] The dual-layer long short-term memory network adopts a dual-layer cascaded LSTM structure: the first layer (first long short-term memory network) has 256 units that output the complete sequence, and the second layer (second long short-term memory network) has 128 units that output the final state.

[0103] Dynamic regularization: A first Dropout layer with a probability of 0.4 is inserted between the first and second Long Short-Term Memory networks, and a second Dropout layer with a probability of 0.3 is inserted between the second Long Short-Term Memory network and the fully connected layer.

[0104] Adaptive learning rate scheduling: Initial learning rate 0.0005, decreasing by 30% every 200 rounds.

[0105] Prediction error compensation: Based on the error distribution characteristics during the testing period, an error threshold alarm and dynamic correction mechanism is established.

[0106] The two-layer Long Short-Term Memory network was trained using the Adam optimizer with a learning rate of 0.0005 and a decay of 30% every 200 epochs. Overfitting was monitored on the validation set.

[0107] ;

[0108] in, For training objective function; For the first The true label of each sample; For the first The model prediction value for each sample; This represents the total number of training samples.

[0109] Calculate MAE, RMSE, The indicators generate a prediction-to-prediction curve, an error histogram, and an actual-to-prediction scatter plot.

[0110] ;

[0111] ;

[0112] ;

[0113] Among them, MAE is the mean absolute error, which measures the average deviation between the predicted value and the actual value; RMSE is the root mean square error, which is more sensitive to larger errors. The coefficient of determination represents the model's ability to explain variations in the data; The mean of the labels; This represents the total number of test samples.

[0114] In another exemplary embodiment of this application, reference is made to... Figure 4 The method described in this application can be executed by a data preprocessing module, a feature engineering module, a standardization module, an LSTM network module, and an evaluation module, comprising steps 101 to 106. The data preprocessing module reads the data and converts kilometer marker strings into numerical values; the feature engineering module constructs lag features, difference features, and moving average features; the standardization module processes the data using robust standardization (median and interquartile range); the LSTM network module constructs a two-layer LSTM structure and adds Dropout to prevent overfitting; and the evaluation module calculates MAE, RMSE, and other parameters. The indicators generate error distribution maps and prediction comparison maps.

[0115] Figure 5 The comparison between the predicted values ​​and actual values ​​of the kilometer marker prediction model is shown. Figure 6The time series of prediction errors for the kilometer marker prediction model is shown. Figure 7 This shows the correlation between the predicted values ​​and actual values ​​from the kilometer marker prediction model. Figures 5-7 It can be seen that the method of this application has a small prediction error, a high degree of correlation between predictions, and a small error between the predicted value and the actual value, thus meeting the required prediction requirements.

[0116] This application utilizes LSTM technology to predict satellite orbit kilometer values, and effectively improves the accuracy of prediction parameters during the prediction process.

[0117] Based on the same inventive concept, this application also provides a satellite orbit kilometer marker parameter prediction device for implementing the satellite orbit kilometer marker parameter prediction method described above. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more satellite orbit kilometer marker parameter prediction device embodiments provided below can be found in the limitations of the satellite orbit kilometer marker parameter prediction method described above, and will not be repeated here.

[0118] In one exemplary embodiment, such as Figure 8 As shown, a satellite orbit kilometer marker parameter prediction device is provided, comprising: a conversion module, a vector projection module, a feature stitching module, a feature merging module, a standard module, a training module, and an application module.

[0119] The system comprises the following modules: a conversion module, which acquires satellite positioning data of multiple key locations on the railway track section obtained by satellite positioning, and converts the latitude and longitude of the data into UTM coordinates; a vector projection module, which projects the UTM coordinates onto the line segments between known kilometer markers on the railway track section to determine the kilometer markers of each key location; a feature stitching module, which stitches multi-time-step lag features onto the satellite positioning data and kilometer markers of each key location to obtain the extended feature matrices of the satellite positioning data and kilometer markers at each key location; and a feature merging module, which determines the difference features and moving average features of each extended feature matrix and merges the extended features. The system generates the final feature matrix using matrices, their difference features, and moving average features. A standard module robustly standardizes each final feature matrix to obtain standard feature matrices for both satellite positioning data and kilometer markers at key locations. A training module uses the satellite feature sequences at each time step of the standard feature matrices for satellite positioning data at key locations as input, and the kilometer markers at each time step of the standard feature matrices for kilometer markers at key locations as labels, to train a two-layer long short-term memory network and obtain a kilometer marker prediction model. An application module inputs the satellite feature sequences of the locations to be measured on the railway track section into the kilometer marker prediction model and outputs the kilometer markers for the locations to be measured on the railway track section.

[0120] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 9 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs in the non-volatile storage media to run. The database stores kilometer markers for locations to be measured on railway track sections. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for predicting satellite orbit kilometer marker parameters.

[0121] Those skilled in the art will understand that Figure 9 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0122] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0123] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0124] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0125] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0126] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0127] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for predicting satellite orbit kilometer marker parameters, characterized in that, include: Acquire satellite positioning data of multiple key locations on the railway track section obtained by satellite positioning, and convert the latitude and longitude in the data into UTM coordinates; Based on the line segments between known kilometer markers on the railway track section, the UTM coordinates are vector-projected onto the line segments to determine the kilometer markers of each key location point; By stitching together the satellite positioning data and kilometer markers at each key location point with the lag features of multiple time steps, the extended feature matrices of the satellite positioning data and kilometer markers at each key location point are obtained. Determine the difference features and moving average features of each extended feature matrix, and generate the final feature matrix from each extended feature matrix and its difference features and moving average features; Robust normalization is performed on each final feature matrix to obtain the standard feature matrices of satellite positioning data and kilometer markers at each key location point; Using the satellite feature sequence of each time step in the standard feature matrix of satellite positioning data at each key location point as input, and the kilometer markers of each time step in the standard feature matrix of each key location point as labels, a two-layer long short-term memory network is trained to obtain the kilometer marker prediction model. The satellite feature sequence of the location point to be measured on the railway track section is input into the kilometer marker prediction model, and the kilometer marker of the location point to be measured on the railway track section is output.

2. The satellite orbit kilometer marker parameter prediction method according to claim 1, characterized in that, Convert the latitude and longitude in this data to UTM coordinates, specifically including: Use longitude from satellite positioning data to determine the UTM partition number; Based on the ellipsoid parameters and the latitude from satellite positioning data, using the formula... and Determine the radius of curvature; where, Let be the radius of curvature. For the semi-major axis of the ellipsoid, For eccentricity, Latitude For ellipsoidal parameters; Based on the radius of curvature, using the formula Determine the east coordinates of the UTM; where, For UTM East coordinates, For eastward pseudo-offset, As a scaling factor, For longitude difference, , Longitude The longitude of the central meridian. The square term of the latitude tangent. , For correction terms related to eccentricity and latitude, ; Based on the radius of curvature, using the formula Determine the north coordinates of the UTM; where, UTM North coordinates The distance on an ellipsoid from the equator to latitude is... UTM North Coordinates The formula is expanded with an even number of terms, and all expansions are calculated using WGS84 ellipsoidal parameters, representing the ellipsoidal distance from the equator to latitude. The calculation formula is the standard meridian arc length formula, expressed in the form of a Fourier series; The UTM partition number, UTM east coordinate, and UTM north coordinate together constitute the UTM coordinate.

3. The satellite orbit kilometer marker parameter prediction method according to claim 1, characterized in that, Based on the line segments between known kilometer markers on the railway track section, the UTM coordinates are vector-projected onto the line segments to determine the kilometer markers of each key location point, specifically including: Determine the line segments between all known kilometer markers on the railway track section; Based on the UTM coordinates of each key location point and the endpoints of each line segment, the formula is used. Determine the projection parameters of the UTM coordinates of each key location point onto each line segment; where, For projection parameters, , The distance from the UTM coordinates of the critical location point to one endpoint of the line segment. , UTM coordinates of key location points Let the coordinates of one endpoint of the line segment be . The distance from the UTM coordinates of the critical location point to the other endpoint of the line segment. , Let the coordinates be the coordinates of the other endpoint of the line segment; Based on the projection parameters, using the formula Determine the projection coordinates of the UTM coordinates of each key location point onto each line segment; where, The coordinates of the projection point; Based on the UTM coordinates and projection point coordinates of each key location point, select the projection point coordinates and line segment corresponding to the minimum projection distance, and use the selected line segment as the projection line segment. If the coordinates of the selected projection point lie on the projection line segment, then use the formula Determine the kilometer markers for each key location; where, For key location kilometer markers, The kilometer marker at one endpoint of the projected line segment. The kilometer marker at the other endpoint of the projected line segment; If the selected projection point coordinates are outside the projection line segment, then the kilometer marker of the endpoint on the projection line segment closest to the projection point coordinates will be determined as the kilometer marker of each key location point.

4. The satellite orbit kilometer marker parameter prediction method according to claim 1, characterized in that, By concatenating satellite positioning data from key locations with lag features from multiple time steps, an extended feature matrix of satellite positioning data at each key location is obtained, specifically including: The satellite positioning data with multiple time steps of lag are used to construct a lag feature vector; By concatenating the satellite positioning data of each key location point with the lag feature vector, an extended feature matrix of the satellite positioning data at each key location point is obtained.

5. The satellite orbit kilometer marker parameter prediction method according to claim 1, characterized in that, Determine the difference features and moving average features of each extended feature matrix, and generate the final feature matrix from each extended feature matrix and its difference features and moving average features. Specifically, this includes: Using formula Calculate the difference features of each extended feature matrix; where, For time step Difference features of the extended feature matrix, For time step The extended feature matrix, For time step The extended feature matrix; Using formula Calculate the moving average characteristic of each extended characteristic matrix; where, For time step Moving average feature of extended feature matrix, For time step The extended feature matrix, The offset within the window. =0,1,2; The final feature matrix is ​​obtained by merging the extended feature matrices with the corresponding difference features and moving average features.

6. The satellite orbit kilometer marker parameter prediction method according to claim 1, characterized in that, Robust normalization is performed on each final feature matrix to obtain the standard feature matrices of satellite positioning data and kilometer markers at each key location point, specifically including: Determine the median and interquartile range of each final feature matrix; Based on the median and the interquartile range, using the formula The final feature matrices are standardized to obtain the standard feature matrices of satellite positioning data and kilometer markers at each key location point; where, For the final characteristic matrix, the th The standardized feature values ​​of each feature For the th in the final feature matrix One characteristic, For the final characteristic matrix, the th The median of the features, For the final characteristic matrix, the th Interquartile range of each feature.

7. The satellite orbit kilometer marker parameter prediction method according to claim 1, characterized in that, The two-layer long short-term memory network consists of: a first long short-term memory network, a first dropout layer, a second long short-term memory network, a second dropout layer, and a fully connected layer connected in sequence; The dual-layer long short-term memory network has an adaptive learning rate.

8. A satellite orbit kilometer marker parameter prediction device, characterized in that, include: The conversion module is used to acquire satellite positioning data of multiple key locations on the railway track section measured by satellite positioning, and convert the latitude and longitude in the data into UTM coordinates; The vector projection module is used to perform vector projection of UTM coordinates on the line segments between known kilometer markers on the railway track section to determine the kilometer markers of each key location point; The feature stitching module is used to stitch together the satellite positioning data and kilometer markers at each key location point with lag features at multiple time steps to obtain the extended feature matrices of the satellite positioning data and kilometer markers at each key location point. The feature merging module is used to determine the difference features and moving average features of each extended feature matrix, and to generate the final feature matrix from each extended feature matrix and its difference features and moving average features. The standard module is used to robustly normalize each final feature matrix to obtain the standard feature matrices of satellite positioning data and kilometer markers at each key location point; The training module is used to train a two-layer long short-term memory network with satellite feature sequences at each time step in the standard feature matrix of satellite positioning data at each key location point as input and kilometer markers at each time step in the standard feature matrix of kilometer markers at each key location point as labels, to obtain a kilometer marker prediction model. The application module is used to input the satellite feature sequence of the location points to be measured on the railway track section into the kilometer marker prediction model and output the kilometer markers of the location points to be measured on the railway track section.

9. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the satellite orbit kilometer marker parameter prediction method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the satellite orbit kilometer marker parameter prediction method as described in any one of claims 1-7.

Images