Multi-source target trajectory fusion method and system based on deep learning

Abstract

The invention relates to the technical field of data processing, in particular to a multi-source target trajectory fusion method and system based on deep learning. Comprising the following steps: constructing a space-time alignment training set containing radar echo data, satellite telemetry data and truth value trajectory data; designing a multi-channel deep learning network architecture which comprises a radar echo data feature extraction channel and a satellite telemetry data feature extraction channel, and establishing dual-channel feature association through an attention mechanism; optimizing network parameters through back propagation to obtain a weight prediction model; radar echo data and satellite telemetry data are received in real time, and space-time alignment of sampling data is achieved through a sliding window mechanism and coordinate system conversion; inputting the preprocessed double-source features into the trained weight prediction model, and generating a radar-telemetering fusion weight coefficient; and performing weighting processing and optimization on the double-source trajectory coordinates, and outputting a fused target motion trajectory. The problem that a traditional single-source track measurement system is insufficient in precision in a complex environment is solved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a multi-source target trajectory fusion method and system based on deep learning. Background Art

[0002] In the prior art, the target trajectory measurement system mainly relies on a single sensor data source, and the measurement accuracy in complex environments has significant limitations. When using satellite telemetry data, it is limited by the transmission conditions of satellite signals. When encountering occlusion, electromagnetic interference, etc., the signal is easily lost or deviated, resulting in incomplete or inaccurate measured trajectory data. Moreover, the time resolution of the satellite is usually low, and it is easy to produce centimeter-level spatial offset due to atmospheric interference, making it difficult to accurately capture the instantaneous posture changes of high-speed maneuvering targets. When using radar echo data, although the radar has a millisecond-level time resolution, it is affected by environmental factors such as multipath effects and ground obstruction. In a complex electromagnetic environment, the target scattering characteristics will be distorted, making it difficult to extract the target scattering characteristics, thereby affecting the judgment of the target position and trajectory. In addition, the detection range and accuracy of the radar also have certain limitations. For the trajectory measurement of long-distance or small targets, its accuracy will be significantly reduced.

[0003] Advanced target trajectory measurement requires the integration of the complementary advantages of radar and satellite, but existing technologies face severe challenges in the deep fusion of heterogeneous data. Since radar echo data is essentially a scattering feature matrix in the time-frequency domain, and satellite telemetry data is a discrete positioning parameter in the geographic coordinate system, there are essential differences between the two in terms of data structure dimension, spatiotemporal reference, and error characteristics. Existing fusion methods cannot establish a correlation model for cross-modal features, and traditional fixed-weight fusion mechanisms are also difficult to adaptively adjust the confidence weights of radar and telemetry data based on dynamic factors such as environmental interference intensity and target maneuvering state. For target trajectory measurement systems with high precision requirements, existing technologies cannot effectively provide accurate, timely and effective trajectory fusion support, making it difficult to meet the high-precision tracking and positioning requirements of targets in complex environments in practical applications, limiting the performance and reliability of trajectory measurement systems in actual operations. Summary of the invention

[0004] In order to solve the problem of insufficient accuracy of traditional single-source trajectory measurement systems in complex environments, the present invention provides a multi-source target trajectory fusion method and system based on deep learning.

[0005] In a first aspect, the present invention provides a multi-source target trajectory fusion method based on deep learning, which adopts the following technical solution: A multi-source target trajectory fusion method based on deep learning, comprising: Construct a spatiotemporal alignment training set containing radar echo data, satellite telemetry data, and true trajectory data; Design a multi-channel deep learning network architecture, including echo data feature extraction channel and telemetry data feature extraction channel, and establish dual-channel feature association through attention mechanism; The network parameters are optimized through back propagation to obtain the weight prediction model; Receive radar echo data and satellite telemetry data in real time, and achieve spatial and temporal alignment of sampled data through sliding window mechanism and coordinate system conversion; The preprocessed dual-source features are input into the trained weight prediction model to generate radar-telemetry fusion weight coefficients; The dual-source trajectory coordinates are weighted and optimized, and the fused target motion trajectory is output.

[0006] Furthermore, the construction of a spatiotemporal alignment training set comprising radar echo data, satellite telemetry data and true value trajectory data comprises: Perform pulse compression processing on the original radar echo data and extract error parameters to construct a feature matrix; Perform coordinate system conversion on satellite telemetry data, extract error parameters, and construct feature matrix; The sliding window mechanism and cubic spline interpolation method are used to align and complete the time axis of the dual-source data and establish millisecond-level synchronized training samples.

[0007] Furthermore, the multi-channel deep learning network architecture is designed, including a radar echo data feature extraction channel and a satellite telemetry data feature extraction channel, and establishing dual-channel feature association through an attention mechanism includes: Radar echo data feature extraction channel, using a three-dimensional convolutional neural network to extract time-space-error features; Satellite telemetry data feature extraction channel, using a bidirectional LSTM network to decode satellite positioning parameter sequences; Cross-attention module, which builds the correlation matrix between radar features and telemetry features containing environmental quality perception; The fully connected output layer generates a weight distribution tensor of dimension [timestamp × spatial coordinate].

[0008] Furthermore, the cross attention module calculates the correlation matrix of the radar feature and the telemetry feature, including: Mapping radar feature vectors and telemetry feature vectors to query space and key value space respectively; Calculate the feature similarity matrix and generate the attention weight distribution through the softmax function; The dual-channel features are weightedly fused to generate a joint feature representation and input into the fully connected layer.

[0009] Furthermore, the method of optimizing network parameters by back propagation to obtain a weight prediction model includes: Initialize network parameters and define the mean square error loss function between the fusion trajectory and the true value trajectory; Use back propagation and stochastic gradient descent algorithms to update network weights; The model overfitting is controlled by early stopping method, and the optimal weight prediction model is saved.

[0010] Furthermore, the real-time reception of radar echo data and satellite telemetry data and the realization of spatiotemporal alignment of sampled data through a sliding window mechanism and coordinate system conversion include: Perform pulse compression processing on the original radar echo data and extract error parameters to construct a feature matrix; Perform coordinate system conversion on satellite telemetry data, extract error parameters, and construct feature matrix; The sliding window mechanism and cubic spline interpolation method are used to align and complete the time axis of the dual-source data.

[0011] Furthermore, the weighted processing and optimization of the dual-source trajectory coordinates and the output of the fused target motion trajectory include: Perform weighted calculation on the radar trajectory point P_r and the telemetry trajectory point P_s: P_fused = α·P_r + β·P_s, where P_fused is the coordinate of the fusion point, α and β are the fusion weight coefficients of radar data and telemetry data respectively; Perform sliding average filtering on the fused trajectory to remove noise and output the optimized target motion trajectory.

[0012] In the second aspect, a multi-source target trajectory fusion system based on deep learning includes: The module for building a spatiotemporal alignment training set is used to resample the radar echo data, satellite telemetry data, and true value trajectory in time axis and coordinate system to generate a spatiotemporal synchronized supervised learning dataset; The model network architecture building module is connected with the spatiotemporal alignment training set building module to build a deep learning model network architecture suitable for multi-source target trajectory fusion; The model training module is connected to the model network architecture construction module, uses the constructed training set to train the constructed network architecture, generates a dynamic weight prediction model, and saves the optimal parameter model; The sampling data spatiotemporal alignment module aligns the multi-source trajectory data acquired in real time, making the data from different sources comparable in time and space dimensions; A fusion weight coefficient generation module is connected to the model training module and the sampling data spatiotemporal alignment module, and is used to input the preprocessed dual-source features into the training model to generate weight coefficients for trajectory fusion; The multi-source trajectory weighted processing module is connected to the fusion weight coefficient generation module and is used to perform linear weighted calculation on the radar trajectory points and the telemetry trajectory points according to the weight coefficients to generate an initial fusion trajectory; The fusion trajectory filtering and smoothing module is connected to the multi-source trajectory weighted processing and smoothing module to perform filtering and smoothing operations on the fusion trajectory after weighted processing to remove noise and fluctuations, making the trajectory smoother and more accurate; The target fusion trajectory output module is connected to the fusion trajectory filtering and smoothing module to output the final optimized target fusion trajectory.

[0013] In summary, the present invention has the following beneficial technical effects: 1. The multi-source target trajectory fusion method based on deep learning proposed in this paper, by constructing a spatiotemporal alignment training set containing radar echoes, satellite telemetry and true value data, and designing a dual-channel deep learning network with a cross-attention mechanism, has achieved the cross-modal dynamic correlation modeling of radar time-frequency features and satellite positioning parameters for the first time. Compared with the traditional fixed weight fusion method, this scheme can adaptively generate the optimal fusion weight coefficient based on dynamic factors such as the target maneuvering state and environmental interference intensity, greatly improving the trajectory positioning accuracy in a complex electromagnetic environment, and effectively solving the feature mismatch problem caused by the dimensional differences of heterogeneous data.

[0014] 2. The sliding window mechanism and coordinate system conversion method adopted by the present invention, combined with the cubic spline interpolation method, resamples the dual-source data in time axis, realizes high-precision spatiotemporal alignment of the sampled data, solves the problem of differences in spatiotemporal benchmarks between different data sources, provides a reliable basis for subsequent data fusion, and improves the accuracy and stability of the entire trajectory fusion process. Whether it is when the target's high-speed maneuvers cause drastic changes in data, or when the data fluctuates abnormally in a complex environment, the accuracy and integrity of the data can be improved, thereby improving the accuracy and stability of the entire trajectory fusion process, so that the fused trajectory can more truly reflect the actual movement of the target.

[0015] 3. The present invention optimizes network parameters through back propagation, defines the mean square error loss function of the fusion trajectory and the true value trajectory, and uses the early stopping method to control the overfitting of the model. The obtained weight prediction model can dynamically adjust the fusion weight according to different data characteristics. Compared with the traditional fixed weight fusion mechanism, the fusion weight can be adjusted more flexibly and dynamically. Whether facing sudden changes in the intensity of environmental interference or complex changes in the maneuvering state of the target, it can respond quickly and adjust the weight distribution to make the fused target motion trajectory more accurate, greatly improving the adaptability and accuracy of the model.

[0016] 4. The cross-attention module of the present invention constructs a joint mapping mechanism of radar features and telemetry features, and establishes a nonlinear association model of dual-source features through query-key space conversion and softmax weight allocation. The module significantly improves the trajectory continuity retention rate in scenarios where a single signal is blocked, significantly enhances the anti-interference ability in complex environments, and provides a strong guarantee for the application of multi-source target trajectory fusion in complex environments, ensuring more accurate acquisition of target motion trajectory information in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flow chart of the method of embodiment 1 of the present invention; Figure 2 is a schematic diagram of the system structure of Embodiment 2 of the present invention; DETAILED DESCRIPTION

[0018] The present invention is further described in detail below in conjunction with the accompanying drawings. Example 1

[0019] Reference Figure 1 , a multi-source target trajectory fusion method based on deep learning in this embodiment includes: S1. Construct a spatiotemporal alignment training set containing radar echo data, satellite telemetry data, and true trajectory data, including: S11. Perform pulse compression processing on the original radar echo data and extract error parameters to construct a feature matrix; The original radar echo signal is subjected to matched filter pulse compression processing to generate a time-frequency two-dimensional feature matrix, and the scattering feature vector is extracted to obtain the target position. During the feature extraction process, the signal-to-noise ratio (SNR) is synchronously extracted as the error feature parameter. By introducing the error feature parameter, the dynamic change of the radar measurement quality can be effectively characterized, providing a quantitative basis for the subsequent weight allocation. The scattering feature vector and the error parameter are spliced ​​into a composite feature vector with the dimension [timestamp × target position × SNR].

[0020] S12. Perform coordinate system conversion on satellite telemetry data and extract error parameters to construct a feature matrix; The conversion formula from WGS84 coordinate system to ENU coordinate system is used to convert the original satellite positioning data to the radar station center coordinate system. During the conversion process, the geometric dilution of precision (DOP) parameters of the satellite system are synchronously extracted as error characteristic parameters to construct a three-dimensional feature matrix with the dimension of [timestamp × target position × DOP]. The DOP parameter includes components such as HDOP and VDOP, which are used to quantify the characteristics of satellite positioning accuracy as the geometric configuration of the constellation changes, and provide an error characterization basis for subsequent feature fusion.

[0021] S13. Use the sliding window mechanism and cubic spline interpolation method to align and fill the time axis of the dual-source data and establish millisecond-level synchronized training samples.

[0022] Set the adaptive window length L=2f_max / f_min, where f_max and f_min are the highest and lowest sampling frequencies of the dual sources, respectively, and use the time axis sliding step Δt=1 / (2f_max) to align the dual source data in the time domain. Then, check whether there are missing values ​​in the radar echo data and satellite telemetry data within the sliding window. If there are missing data, use cubic spline interpolation to fill them. This process helps to reduce the errors caused by environmental factors (such as high radar signal-to-noise ratio or weak satellite signal) by filling the missing data in the window while ensuring data smoothness. It also solves the problem of inconsistent radar and satellite sampling frequencies, providing a high-quality data set for subsequent model training.

[0023] S2. Design a multi-channel deep learning network architecture, including echo data feature extraction channel and telemetry data feature extraction channel, and establish dual-channel feature association through attention mechanism, including: S21. Radar echo data feature extraction channel, using 3D convolutional neural network to extract time-space-error features; The input layer receives a composite feature vector of dimension [timestamp × target position × SNR]. The first layer convolution kernel is designed as a (3×3×16) time-space-error three-dimensional convolution, which can fully extract multi-dimensional features. The second layer uses a 3×3 dilated convolution, which can expand the receptive field and capture a wider range of feature information without increasing the number of parameters. The last layer uses a 1×1 convolution for feature compression, removes redundant information, and refines key features. Each layer of convolution is followed by an attention gating module, which dynamically adjusts the feature weights according to the SNR value of each time slice.

[0024] S22. Satellite telemetry data feature extraction channel, using a bidirectional LSTM network to decode satellite positioning parameter sequences; A bidirectional LSTM network with 64 hidden units is constructed, and the input sequence is a triple feature of [timestamp × target position × DOP]. During the time back propagation process, a tanh activated scaling factor is applied to the DOP parameter. When DOP>3, the output feature is attenuated by 30%, and when DOP<2, the feature is enhanced by 20%. Through this DOP-sensitive feature modulation, the network can autonomously suppress the influence of positioning data during high error periods.

[0025] S23. Cross-attention module, which builds the correlation matrix between radar features and telemetry features with environmental quality perception, including: S231. Mapping the radar feature vector and the telemetry feature vector to the query space and the key value space respectively; The bilinear mapping matrices W_q∈R^(d×d) and W_k∈R^(d×d) are designed, where d is the feature dimension. The radar feature F_r generates a query vector through F_q=F_rW_q, and the satellite feature F_s generates a key vector through F_k=F_sW_k. In particular, the error parameter SNR of the radar echo data and the error parameter DOP of the satellite telemetry data are retained as additional feature channels during the mapping process, so that the attention mechanism can synchronously perceive the environmental quality parameters.

[0026] S232. Calculate the feature similarity matrix and generate the attention weight distribution through the softmax function; The similarity between the radar feature vector and the telemetry feature vector is calculated by dot product to obtain the feature similarity matrix. Each element in this matrix represents the similarity between the corresponding element in the radar feature vector and the telemetry feature vector. Then, the feature similarity matrix is ​​input into the softmax function, which normalizes each element in the similarity matrix so that the sum of all elements is 1, thereby generating the attention weight distribution.

[0027] S233. Perform weighted fusion on the dual-channel features to generate a joint feature representation and input it into the fully connected layer.

[0028] According to the generated attention weight distribution, the radar feature vector and the telemetry feature vector are weighted fused to obtain a joint feature representation. This joint feature representation combines the information of radar features and telemetry features, effectively integrating the advantages of dual-source data. The joint feature representation is then input into the fully connected layer for further processing and feature integration.

[0029] S24. Fully connected output layer, generating a weight distribution tensor of dimension [timestamp × spatial coordinate].

[0030] The fully connected layer further linearly transforms and integrates the joint features through a series of weight matrix and bias vector operations. Finally, it outputs a weight distribution tensor with the dimension [timestamp × spatial coordinate]. Each element in this tensor represents the weight distribution of radar data and telemetry data in the fusion process at different timestamps and spatial coordinates.

[0031] S3. Optimize network parameters through back propagation to obtain the weight prediction model, including: S31. Initialize network parameters and define the mean square error loss function between the fusion trajectory and the true value trajectory; During the training process, the parameters of the deep learning network are first initialized, and the mean square error (MSE) is used as the loss function. This loss function can measure the gap between the predicted trajectory and the true trajectory, further optimizing the performance of the network. In addition, the loss function also includes considerations of the impact of environmental data, allowing the network to adaptively adjust during training and reduce the propagation of errors.

[0032] S32. Update network weights using back propagation and stochastic gradient descent algorithms; During the model training process, the dual-source feature matrix constructed in the training set is input into the network. After feature extraction of each channel, correlation calculation of the cross-attention module, and processing of the fully connected layer, the prediction result of the fusion trajectory is obtained. Then, according to the defined mean square error loss function, the error between the prediction result and the true value trajectory is calculated. Using the back propagation algorithm, the error is propagated to each layer of the network in turn in the opposite direction of the network forward propagation, and the gradient of the parameters of each layer is calculated. Then, the weight parameters in the network are updated through the stochastic gradient descent algorithm, so that the value of the loss function gradually decreases.

[0033] S33. Control model overfitting through early stopping method and save the optimal weight prediction model.

[0034] The specific approach is to divide the training data set into a training set and a validation set. After each training iteration, the validation set is used to evaluate the model and the mean square error loss on the validation set is calculated. When it is found that the loss on the validation set does not decrease or increases for several consecutive times (such as 3 times), it is considered that the model may be overfitting. At this time, stop the model training in time and save the weight parameters obtained from the current training. The model corresponding to these parameters is the optimal weight prediction model, which ensures the reliability and generalization ability of the model in practical applications.

[0035] S4. Receive radar echo data and satellite telemetry data in real time, and achieve spatial and temporal alignment of sampled data through sliding window mechanism and coordinate system conversion, including: S41. Perform pulse compression processing on the original radar echo data and extract error parameters to construct a feature matrix; Similar to step S11, the original radar echo signal is subjected to matched filter pulse compression processing to generate a time-frequency two-dimensional feature matrix, and the scattering feature vector is extracted to obtain the target position. During the feature extraction process, the signal-to-noise ratio (SNR) is synchronously extracted as an error feature parameter. The scattering feature vector and the error parameter are then concatenated into a composite feature vector with the dimension [timestamp × target position × SNR].

[0036] S42. Perform coordinate system conversion on satellite telemetry data and extract error parameters to construct a feature matrix; Similar to step S12, the conversion formula from WGS84 coordinate system to ENU coordinate system is used to convert the original satellite positioning data to the radar station center coordinate system. During the conversion process, the geometric dilution of precision (DOP) parameters of the satellite system are synchronously extracted as error characteristic parameters to construct a three-dimensional feature matrix with the dimension [timestamp × target position × DOP].

[0037] S43. The sliding window mechanism and cubic spline interpolation method are used to align and complete the time axis of the dual-source data.

[0038] Similar to step S13, the adaptive window length L=2f_max / f_min is set, where f_max and f_min are the highest and lowest sampling frequencies of the dual sources, respectively, and the time axis sliding step Δt=1 / (2f_max) is used to align the dual source data in the time domain. Then, within the sliding window, check whether there are missing values ​​in the radar echo data and satellite telemetry data. If there are missing data, use the cubic spline interpolation method to fill them.

[0039] S5. Input the preprocessed dual-source features into the trained weight prediction model to generate radar-telemetry fusion weight coefficients.

[0040] The radar echo data, a three-dimensional feature matrix in the format of [timestamp × target position × DOP], and the satellite telemetry data, a three-dimensional feature matrix in the format of [timestamp × target position × SNR], are input into the trained weight prediction model. After calculation by the deep learning network, the model will generate fusion weight coefficients α and β according to the characteristics of each data source. By adjusting these coefficients, the model can dynamically optimize the fusion ratio of radar and satellite data according to the environment and target status.

[0041] S6. Perform weighted processing and optimization on the dual-source trajectory coordinates, and output the fused target motion trajectory, including: S61. Perform weighted calculation on the radar trajectory point P_r and the telemetry trajectory point P_s: P_fused = α·P_r +β·P_s, where P_fused is the coordinate of the fusion point, and α and β are the fusion weight coefficients of the radar data and the telemetry data, respectively; In this step, the radar track point P_r and the satellite telemetry track point P_s are weighted according to the previously calculated weight coefficients α and β. Specifically, the final coordinates of each track point P_fused are the result of the fusion of radar and satellite data, and the contributions of the two data sources are adjusted by the fusion weights α and β. This weighting process can be dynamically adjusted according to different environmental factors to ensure that in practical applications, the fusion of radar data and satellite data can accurately reflect the motion trajectory of the target.

[0042] S62. Perform sliding average filtering on the fused trajectory to remove noise and output the optimized target motion trajectory.

[0043] Adopt Savitzky-Golay filter with adaptive window length, window size , where SNR_avg is the average radar signal-to-noise ratio in the current window, and DOP_avg is the average satellite precision factor. The smoothing window is automatically increased to 15 points in low signal-to-noise ratio or high DOP periods, and reduced to 5 points in high signal-to-noise ratio and low DOP periods, to achieve environment-adaptive trajectory optimization. After filtering, the output target motion trajectory is smoother and more accurate, and can provide more reliable target position prediction, especially in real-time dynamic environment, the tracking of target motion has stronger robustness.

[0044] This embodiment provides a multi-source target trajectory fusion method based on deep learning. The method aims at the background that the traditional single-source trajectory measurement system has insufficient accuracy in complex environments and the existing multi-source fusion technology is difficult to effectively fuse radar echoes and satellite telemetry data. In order to accurately generate the target motion trajectory under complex and changeable electromagnetic environments and various interference conditions, a spatiotemporal alignment training set is constructed to perform spatiotemporal alignment processing on radar echo data and satellite telemetry data, and high-precision spatiotemporal synchronization is achieved by using a cubic spline interpolation method. A multi-channel deep learning network architecture is designed, and dual-channel feature association is established with the help of an attention mechanism, a cross-modal data association model is established, and feature information of the two data sources is deeply mined. The trained weight prediction model is applied to real-time data processing, and the radar echo data and satellite telemetry data received in real time are spatiotemporally aligned through a sliding window mechanism and coordinate system conversion. The model is then input to generate a fusion weight coefficient, and then the dual-source trajectory coordinates are weighted and optimized, and finally a high-precision fused target motion trajectory is output, which significantly improves the accuracy and stability of trajectory measurement. This method can also efficiently and accurately fuse radar and satellite data from different sources and with different characteristics in actual scenarios with high requirements for trajectory measurement accuracy, thereby providing accurate, timely and reliable target trajectory data for these fields with high precision requirements, and effectively improving the reliability of task execution and equipment effectiveness in related fields. Example 2

[0045] Reference Figure 2 , this embodiment provides a multi-source target trajectory fusion system based on deep learning, including: The module for building a spatiotemporal alignment training set is used to resample the radar echo data, satellite telemetry data, and true value trajectory in time axis and coordinate system to generate a spatiotemporal synchronized supervised learning dataset; The model network architecture building module is connected with the spatiotemporal alignment training set building module to build a deep learning model network architecture suitable for multi-source target trajectory fusion; The model training module is connected to the model network architecture construction module, uses the constructed training set to train the constructed network architecture, generates a dynamic weight prediction model, and saves the optimal parameter model; The sampling data spatiotemporal alignment module aligns the multi-source trajectory data acquired in real time, making the data from different sources comparable in time and space dimensions; A fusion weight coefficient generation module is connected to the model training module and the sampling data spatiotemporal alignment module, and is used to input the preprocessed dual-source features into the training model to generate weight coefficients for trajectory fusion; The multi-source trajectory weighted processing module is connected to the fusion weight coefficient generation module and is used to perform linear weighted calculation on the radar trajectory points and the telemetry trajectory points according to the weight coefficients to generate an initial fusion trajectory; The fusion trajectory filtering and smoothing module is connected to the multi-source trajectory weighted processing and smoothing module to perform filtering and smoothing operations on the fusion trajectory after weighted processing to remove noise and fluctuations, making the trajectory smoother and more accurate; The target fusion trajectory output module is connected to the fusion trajectory filtering and smoothing module to output the final optimized target fusion trajectory.

[0046] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-source target trajectory fusion method based on deep learning, characterized in that: include: Construct a spatiotemporal alignment training set containing radar echo data, satellite telemetry data, and true trajectory data; Design a multi-channel deep learning network architecture, including echo data feature extraction channel and telemetry data feature extraction channel, and establish dual-channel feature association through attention mechanism; The network parameters are optimized through back propagation to obtain the weight prediction model; Receive radar echo data and satellite telemetry data in real time, and achieve spatial and temporal alignment of sampled data through sliding window mechanism and coordinate system conversion; The preprocessed dual-source features are input into the trained weight prediction model to generate radar-telemetry fusion weight coefficients; The dual-source trajectory coordinates are weighted and optimized, and the fused target motion trajectory is output.

2. According to a multi-source target trajectory fusion method based on deep learning according to claim 1, it is characterized in that: The construction of a spatiotemporal alignment training set containing radar echo data, satellite telemetry data, and true value trajectory data includes: Perform pulse compression processing on the original radar echo data and extract error parameters to construct a feature matrix; Perform coordinate system conversion on satellite telemetry data, extract error parameters, and construct feature matrix; The sliding window mechanism and cubic spline interpolation method are used to align and complete the time axis of the dual-source data and establish millisecond-level synchronized training samples.

3. The multi-source target trajectory fusion method based on deep learning according to claim 1 is characterized in that: The multi-channel deep learning network architecture is designed, including a radar echo data feature extraction channel and a satellite telemetry data feature extraction channel, and dual-channel feature association is established through an attention mechanism, including: Radar echo data feature extraction channel, using a three-dimensional convolutional neural network to extract time-space-error features; Satellite telemetry data feature extraction channel, using a bidirectional LSTM network to decode satellite positioning parameter sequences; Cross-attention module, which builds the correlation matrix between radar features and telemetry features containing environmental quality perception; The fully connected output layer generates a weight distribution tensor of dimension [timestamp × spatial coordinate].

4. The multi-source target trajectory fusion method based on deep learning according to claim 3 is characterized in that: The cross attention module calculates the correlation matrix between radar features and telemetry features, including: Mapping radar feature vectors and telemetry feature vectors to query space and key value space respectively; Calculate the feature similarity matrix and generate the attention weight distribution through the softmax function; The dual-channel features are weightedly fused to generate a joint feature representation and input into the fully connected layer.

5. The multi-source target trajectory fusion method based on deep learning according to claim 1 is characterized in that: The method of optimizing the network parameters by back propagation to obtain the weight prediction model includes: Initialize network parameters and define the mean square error loss function between the fusion trajectory and the true value trajectory; Use back propagation and stochastic gradient descent algorithms to update network weights; The model overfitting is controlled by early stopping method, and the optimal weight prediction model is saved.

6. The multi-source target trajectory fusion method based on deep learning according to claim 1 is characterized in that: The real-time reception of radar echo data and satellite telemetry data and the realization of spatiotemporal alignment of the sampled data through a sliding window mechanism and coordinate system conversion include: Perform pulse compression processing on the original radar echo data and extract error parameters to construct a feature matrix; Perform coordinate system conversion on satellite telemetry data, extract error parameters, and construct feature matrix; The sliding window mechanism and cubic spline interpolation method are used to align and complete the time axis of the dual-source data.

7. The multi-source target trajectory fusion method based on deep learning according to claim 1 is characterized in that: The weighted processing and optimization of the dual-source trajectory coordinates and the output of the fused target motion trajectory include: Perform weighted calculation on the radar trajectory point P_r and the telemetry trajectory point P_s: P_fused = α·P_r + β·P_s, where P_fused is the coordinate of the fusion point, α and β are the fusion weight coefficients of radar data and telemetry data respectively; Perform sliding average filtering on the fused trajectory to remove noise and output the optimized target motion trajectory.

8. A multi-source target trajectory fusion system based on deep learning, characterized in that: The multi-source target trajectory fusion system comprises: The module for building a spatiotemporal alignment training set is used to resample the radar echo data, satellite telemetry data, and true value trajectory in time axis and coordinate system to generate a spatiotemporal synchronized supervised learning dataset; The model network architecture building module is connected with the spatiotemporal alignment training set building module to build a deep learning model network architecture suitable for multi-source target trajectory fusion; The model training module is connected to the model network architecture construction module, uses the constructed training set to train the constructed network architecture, generates a dynamic weight prediction model, and saves the optimal parameter model; The sampling data spatiotemporal alignment module aligns the multi-source trajectory data acquired in real time, making the data from different sources comparable in time and space dimensions; A fusion weight coefficient generation module is connected to the model training module and the sampling data spatiotemporal alignment module, and is used to input the preprocessed dual-source features into the training model to generate weight coefficients for trajectory fusion; The multi-source trajectory weighted processing module is connected to the fusion weight coefficient generation module and is used to perform linear weighted calculation on the radar trajectory points and the telemetry trajectory points according to the weight coefficients to generate an initial fusion trajectory; The fusion trajectory filtering and smoothing module is connected to the multi-source trajectory weighted processing and smoothing module to perform filtering and smoothing operations on the fusion trajectory after weighted processing to remove noise and fluctuations, making the trajectory smoother and more accurate; The target fusion trajectory output module is connected to the fusion trajectory filtering and smoothing module to output the final optimized target fusion trajectory.

Images