Interrupted track association method and device for comparing track motion characteristics, and electronic equipment

Abstract

The embodiment of the invention discloses an interrupted track association method and device for comparing track motion characteristics and electronic equipment. A specific embodiment of the method comprises the following steps: acquiring a new track group; selecting a new track meeting a preset screening condition with the historical track from the new track group to obtain a target track group; inputting the target track group and the historical track into a preset track association neural network to generate a track association probability value sequence; in response to determining that the maximum track association probability value in the track association probability value sequence is smaller than the preset association probability, recording the accumulated interruption duration of the historical track; and in response to determining that the accumulated interruption duration is less than a preset interruption duration threshold, updating the new track group, and executing the track association step again. According to the embodiment, the accuracy of track association interruption can be improved.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the fields of computer technology and track association, and more particularly to a method, device, and electronic device for associating interrupted tracks by comparing track motion characteristics. Background Art

[0002] As a primary situational awareness device, radar's mission is to stably and continuously track targets in the air. However, due to fluctuations in the target's radar cross section, large-angle maneuvers, and Doppler blind spots, radar often experiences tracking interruptions during target detection, leading to track fragmentation and severely impacting air traffic controllers' decision-making efficiency.

[0003] Current technologies for correlating interrupted tracks are primarily categorized into traditional methods and deep learning approaches. Traditional methods include predictive Kalman filter correlation and interactive multi-model correlation, as well as fuzzy track correlation based on similarity metrics. These methods require prior knowledge and assume target motion models, consuming significant manpower and resources for parameter adjustment and model selection, resulting in low practical value. Deep learning methods avoid the complex model selection and parameter adjustment process. Existing technologies include the twin contrastive neural network (TSADCNN), deep temporal contrastive methods, generative correlation methods, and graph representation correlation methods. While these methods have achieved some progress, they still have significant shortcomings: TSADCNN and graph representation methods only implement interaction between feature modules through loss functions; generative methods lack consideration of temporal features; and deep temporal contrastive methods simply extract features through an encoder structure and then perform simple comparisons, failing to fully exploit the spatiotemporal correlations between new and historical tracks.

[0004] Traditional methods rely heavily on prior assumptions and manual parameter adjustment, making them difficult to adapt to the complex and ever-changing real-world environments. Existing deep learning methods fail to adequately consider the interaction between new and historical track features, either employing overly simplistic methods (using only loss functions) or completely ignoring temporal features. Most methods employ symmetric network structures to process sequences of new and historical tracks of unequal length, resulting in inefficient feature extraction. Furthermore, the accuracy of association in scenarios with long interruptions and multiple targets still requires significant improvement. These shortcomings directly impact the accuracy and real-time nature of interrupted track association, hindering air traffic controllers' understanding and decision-making regarding the situation of aerial targets.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the inventive concept and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0006] The content of this disclosure is used to briefly introduce concepts that will be described in detail in the detailed description section below. The content of this disclosure is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0007] Some embodiments of the present disclosure provide a method, device, and electronic device for associating interrupted tracks by comparing track motion characteristics to solve one or more of the technical problems mentioned in the above background technology section.

[0008] In a first aspect, some embodiments of the present disclosure provide a method for associating interrupted tracks by comparing track motion characteristics, the method comprising: in response to detecting a historical track interruption, obtaining a new track group; for the above-mentioned historical tracks and the above-mentioned new track group, performing the following track association steps: selecting new tracks that meet preset screening conditions with the above-mentioned historical tracks from the above-mentioned new track group to obtain a target track group; inputting the above-mentioned target track group and the above-mentioned historical tracks into a preset track association neural network to generate a track association probability value sequence, wherein the above-mentioned track association neural network is used to extract motion characteristics of the new track and the historical track for track association, each track association probability value corresponds to a target track, and the above-mentioned track association neural network includes: a preprocessing module, a track feature extraction module, an auxiliary attention module and a classifier module; in response to determining that the maximum track association probability value in the above-mentioned track association probability value sequence is less than the preset association probability, recording the cumulative interruption duration of the above-mentioned historical tracks; in response to determining that the above-mentioned cumulative interruption duration is less than the preset interruption duration threshold, updating the above-mentioned new track group, and performing the above-mentioned track association step again.

[0009] In a second aspect, some embodiments of the present disclosure provide an interrupted track association device for comparing track motion characteristics, the device comprising: an acquisition unit configured to acquire a new track group in response to detecting a historical track interruption; a track association unit configured to perform the following track association steps for the historical track and the new track group: selecting new tracks that meet preset screening conditions with the historical track from the new track group to obtain a target track group; inputting the target track group and the historical track into a preset track association neural network to generate a track association probability value sequence, wherein The track association neural network is used to extract motion features of new tracks and historical tracks for track association, where each track association probability value corresponds to a target track. The track association neural network includes: a preprocessing module, a track feature extraction module, an auxiliary attention module, and a classifier module. In response to determining that the maximum track association probability value in the track association probability value sequence is less than a preset association probability, the cumulative interruption duration of the historical track is recorded. In response to determining that the cumulative interruption duration is less than a preset interruption duration threshold, the new track group is updated, and the track association step is performed again.

[0010] In a third aspect, some embodiments of the present disclosure provide an electronic device comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0011] In a fourth aspect, some embodiments of the present disclosure provide a computer-readable medium having a computer program stored thereon, wherein when the program is executed by a processor, the method described in any implementation of the first aspect is implemented.

[0012] This paper proposes a method for correlating interrupted tracks based on motion feature comparison. Through an innovative network architecture design (preprocessing module, track feature extraction module, auxiliary attention module, and classifier module), it addresses key challenges in the existing technology. The core of this method lies in the construction of an asymmetric deep neural network. Through multi-module collaboration, this method enables in-depth extraction and effective interaction of the spatiotemporal features of new and historical tracks, significantly improving the accuracy of correlating interrupted tracks. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The above and other features, advantages, and aspects of the various embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that components and elements are not necessarily drawn to scale.

[0014] Figure 1is a flowchart of some embodiments of the obstacle information generation method according to the present disclosure;

[0015] Figure 2 It is a schematic diagram of the track preprocessing module;

[0016] Figure 3 It is a schematic diagram of the track feature extraction module;

[0017] Figure 4 is a schematic diagram of the assistive attention module;

[0018] Figure 5 It is a schematic diagram of the overall structure of the track association neural network;

[0019] Figure 6 is a schematic structural diagram of some embodiments of the obstacle information generating device according to the present disclosure;

[0020] Figure 7 It is a structural diagram of an electronic device suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION

[0021] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of protection of the present disclosure.

[0022] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other.

[0023] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0024] It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, they should be understood as "one or more".

[0025] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0026] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0027] Figure 1 The flowchart 100 of some embodiments of the method for associating interrupted tracks by comparing track motion characteristics according to the present disclosure is shown. The method for associating interrupted tracks by comparing track motion characteristics comprises the following steps:

[0028] Step 101: In response to detecting that a historical track is interrupted, a new track group is obtained.

[0029] In some embodiments, an executing entity of the interrupted track association method for comparing track motion characteristics (eg, a radar device deployed with a track association neural network) acquires a new track group through the radar device in response to detecting a historical track interruption.

[0030] Specifically, during radar detection, a target (such as an aircraft) is continuously tracked, and the target is interrupted (i.e., the track is disconnected) due to reasons such as Doppler blind spots. After a period of interruption, multiple new tracks appear. Thus, a new track group can be obtained.

[0031] In practice, track interruption correlation is a radar target tracking technique used to address target track interruptions caused by target maneuvers, obstructions, or other interference. During radar detection, a target's track may be interrupted due to a sudden turn, obstruction by an obstacle, or signal interference. Following an interruption, the radar system may detect new tracks (i.e., newly emerged target points). Track interruption correlation is the task of matching these newly emerged tracks with previously interrupted historical tracks to determine whether they belong to the same target.

[0032] Step 102: For the historical track and the new track group, perform the following track association steps:

[0033] Step 1021 , selecting new tracks that meet preset screening conditions with historical tracks from the new track group to obtain a target track group.

[0034] In some embodiments, the execution entity may select new tracks from the new track group that meet preset screening conditions with the historical tracks to obtain a target track group.

[0035] In some optional implementations of some embodiments, the execution entity selects new tracks from the new track group that meet preset screening conditions with the historical tracks to obtain a target track group, including:

[0036] The first step is to determine the interruption time, interruption end coordinates, and corresponding target interruption speed for the historical track, as well as the starting coordinates for each new track in the new track group. The interruption time can be the moment the historical track was disconnected. The interruption end coordinates can be the location of the historical track at the time of interruption. The target interruption speed can be the speed of the aircraft at the time of interruption for the historical track.

[0037] The second step is to determine whether each new track in the new track group meets the preset screening condition based on the interruption time point, the interruption end coordinates, and the target interruption speed value. The screening condition is that the distance between the starting coordinates of the new track and the interruption end coordinates is less than or equal to the track distance. The track distance is generated by the interruption time point, the target interruption speed value, and a preset speed difference. The following formula can be used to determine whether each new track in the new track group meets the preset screening condition:

[0038] d es ≤t d ·(v e +Δv)

[0039] Among them, d es Indicates the distance between the last point of the historical track (interruption end coordinates) and the starting point coordinates of the new track, t d represents the track interruption time, v e represents the target speed when the historical track is interrupted (target interruption speed value), and Δv represents the allowable speed difference (preset speed difference). d ·(v e +Δv) is the distance threshold used to filter d es New tracks that are less than this distance threshold.

[0040] In the third step, the new tracks that meet the preset screening conditions are determined as target tracks to obtain a target track group.

[0041] Step 1022: Input the target track group and the historical tracks into a preset track association neural network to generate a track association probability value sequence.

[0042] In some embodiments, the execution entity may input the target track group and the historical tracks into a preset track association neural network to generate a sequence of track association probability values. The track association neural network may be used to extract motion features of the new track and the historical tracks for track association, with each track association probability value corresponding to a target track. The track association neural network includes a preprocessing module, a track feature extraction module, an auxiliary attention module, and a classifier module.

[0043] In some optional implementations of some embodiments, the execution entity inputs the target track group and the historical track into a preset track association neural network to generate a track association probability value sequence, including:

[0044] For each target track in the target track group above, perform the following steps:

[0045] In the first step, the target track and the historical track are input into the pre-processing module in the track association neural network for encoding processing to obtain the historical route coding features and the new route coding features.

[0046] For example, see Figure 2 Schematic diagram of the track preprocessing module shown.

[0047] Here, the preprocessing module includes using a multi-layer perceptron to embed the 3D track vector into a high-dimensional space to capture the relationship between channels. The 3D track vector includes the horizontal coordinate, vertical coordinate and relative time point of the track. Sinusoidal position coding is applied to add position information to the sequence. Thus, the historical route coding features and the new route coding features are obtained. As a result, any position coding can be expressed as a linear function of the previous position, which is conducive to capturing long sequence dependencies. Specifically, the three-dimensional track vector is embedded into a high-dimensional space using a multi-layer perceptron, and the dimension after the multi-layer perceptron is increased to 64 dimensions. Taking into account the large number of Encoder structures used in the Transformer, and the lack of the Transformer model's ability to perceive the order of the input sequence (i.e., permutation invariance), the introduction of position coding explicitly adds position information to each element, enabling the model to distinguish elements in different positions.

[0048] In the second step, the above-mentioned auxiliary attention module and track feature extraction module are used to extract the features of the above-mentioned historical route coding features and new route coding to obtain the historical route feature vector and the new route feature vector.

[0049] For example, see Figure 3 Schematic diagram of the track feature extraction module shown.

[0050] Here, the track feature extraction module is composed of multiple identical layers stacked together, each layer contains a multi-head self-attention sublayer and a feedforward neural network sublayer, and each layer is equipped with a residual connection and a normalization layer. Multi-head self-attention captures sequence relationships from different subspaces by running multiple attention heads in parallel. The feedforward neural network provides nonlinear transformation capabilities. The track feature extraction module also includes 4 identical encoder (device) layers. Deep feature extraction is performed through 4 identical encoder layers. Then, each historical route feature vector and new route feature vector are flattened separately through the flattening layer. Finally, the historical route feature vector and the new route feature vector are obtained through dimensionality reduction processing using a multi-layer perceptron. In practice, the training of learnable network parameters in the track feature extraction module and the preprocessing module is mainly dominated by the contrast loss function.

[0051] As yet another example, see Figure 4 Schematic diagram of the assisted attention module shown.

[0052] Here, the assisted attention module consists of a Bidirectional Long Short-Term Memory (BiLSTM), a self-attention layer, and a feedforward layer (feedforward neural network). The preprocessed features of the historical track are first captured using the BiLSTM to capture bidirectional temporal dependencies. The self-attention layer then highlights key time step information. Finally, a residual connection and a feedforward neural network are used to calculate the historical track's assisted feature vector. The historical track's assisted feature vector serves as the key and value vectors and is input into the attention calculation. The query vector required for the attention calculation comes from the preprocessed feature vector of the new track. The key and value vectors generated by the historical track interact with the query vector of the new track through the attention mechanism. The feature vector of the new track can determine the importance of different positions in its own sequence based on the temporal characteristics of the historical track, allowing track feature extraction to focus more on important parts and ignore unimportant ones.

[0053] In the third step, the historical route feature vectors and the new route feature vectors are classified by the classifier module to generate a track association probability value.

[0054] Specifically, the classifier module consists of a feature concatenation layer, a fully connected layer and a Softmax function. The track vectors of the new track and the historical track after track feature extraction (historical route feature vector and new route feature vector) are directly connected along the channel dimension by the feature connection layer (Feature Concatenate). Then it is input into a fully connected layer (Fullyconnected Layer) with an output dimension of 2. The two output values ​​are respectively represented as: the degree to which the new track and the historical track are the same target (associated), and the degree to which the new track and the historical track are different targets (unassociated). Finally, the output vector is normalized into a probability distribution vector through the Softmax function, so that the final two outputs of the network respectively represent: the probability of track association and the probability of track non-association. The training of the learnable parameters of the classifier module is mainly dominated by the cross entropy loss.

[0055] For example, see Figure 5 The overall structure of the track association neural network is shown in the figure. In the entire network structure diagram, the lower branch (the branch for new route input) is only used to process the feature vector of the new track. Assisted attention does not fuse the features of historical and new tracks, but rather uses the features of historical tracks to assist in expressing the features of the new track.

[0056] In practice, the loss function design of this application combines contrast loss and cross entropy loss. Contrast loss is applied after the new and old track feature extraction module to calculate the feature similarity of the new and old tracks (i.e., new tracks and historical tracks), ensuring that the track features of the same target are close in space and the features of different targets are far apart. Therefore, the loss function should be designed to reduce the difference between the new and old track extraction features of the same target and increase the difference between the new and old track features of different targets. Contrast loss is expressed as:

[0057]

[0058] Where D represents the Euclidean distance between the feature vectors output by the new and old track feature extraction modules (the historical route feature vector and the new route feature vector). Margin>0 represents the edge distance, which is the shortest distance between two track segments belonging to different targets in high-dimensional space. Y + Represents the network prediction label. Loss C Represents the contrast loss value.

[0059] Here, the cross entropy loss is mainly used to guide the training of the classifier module. The auxiliary classifier module converts the distance information between the new and old track feature vectors into the probability information of whether they are associated, and optimizes the classification decision boundary. Its formula can be:

[0060] Loss BCE =-[Y + log(p(Y+ =1))+(1-Y + )log(1-p(Y + =1))]

[0061] Among them, Loss BCE Represents the cross entropy loss value. The label value is 1 or 0. A value of 1 indicates that the new and old tracks are associated, and 0 indicates that they are not associated. p(Y + =1) indicates the probability that the network predicts the label to be 1.

[0062] Then, the above contrast loss value and cross entropy loss value can be weighted summed by the following formula:

[0063]

[0064] Among them, Loss represents the total loss value.

[0065] In addition, the following parameter configurations are used during neural network training in this application: the multilayer perceptron embedding dimension is 64, the hidden layer dimension is 256, and the position encoding dimension is the same as the embedding dimension; the number of self-attention heads is 8, and the hidden layer dimension of the feedforward neural network is 256. The number of BiLSTM hidden units is 64. The hidden layer dimension of the multilayer perceptron after dimensionality reduction is 128, and the output dimension is 32. The edge distance in the contrastive loss is set to 15.0; training uses the Adam optimizer with an initial learning rate of 0.0001 and a batch size of 128. The training dataset must use the actual detection tracks of the radar to be deployed and the simulated tracks of the radar. The tracks are processed at different time intervals and different new track lengths to obtain positive sample data. The new and old tracks of different targets are randomly combined to obtain negative sample data. The positive and negative samples in the dataset must be balanced.

[0066] In practice, the key innovations of this application include the following aspects: (1) Asymmetric network architecture design, which abandons the traditional symmetric dual network structure and adopts a differentiated feature extraction path based on the characteristics of different lengths of the new and old tracks. The old track branch adds an auxiliary attention module, which is more in line with the needs of actual application scenarios. (2) Assisted attention mechanism, after preprocessing the old track vector, the temporal features of the old track are extracted through BiLSTM and self-attention, and the temporal expression of the new track is affected by attention calculation in the form of key-value pairs. Since the vectors output by the new and old tracks after passing through the feature extraction module are used as a measure of similarity comparison, the interaction between the new and old tracks will affect the training of the feature extraction module if it is placed after feature extraction. Moreover, the old track only needs to affect the temporal features of the new track. There is no need to over-extract all the features of the old track. Using a single layer of BiLSTM and self-attention is sufficient, avoiding the increase in network complexity caused by stacking the encoder structure multiple times, while achieving effective interaction between the new and old track features, solving the problem of insufficient feature interaction in existing methods. (3) A multi-layer feature processing process, including a systematic combination of embedding dimension expansion, position encoding, multi-head self-attention, residual connection, layer normalization and other technologies, fully extracts the spatiotemporal features of the track. (4) A dual loss function design, combining contrast loss and cross entropy loss, ensures the discriminability of the feature space and optimizes the classification performance.

[0067] Specifically, the technical effects brought about by this application are mainly reflected in: (1) significantly improving the accuracy of interrupted track association, with the correct association rate reaching 96.4%-99.2% under different interruption times (4-20 seconds), and maintaining an accuracy of 95.2% in multi-target scenarios (50 targets). (2) enhancing the adaptability to long interruption times and complex multi-target scenarios, and effectively utilizing the rich temporal information of old tracks by assisting the attention mechanism. (3) reducing the algorithm's dependence on manual parameter adjustment and empirical knowledge, and realizing end-to-end automated track association. (4) improving the real-time performance of the algorithm, and optimizing the computational efficiency through asymmetric network design and feature dimensionality reduction strategies. (5) with good scalability, the modular design of the network facilitates adjustment and optimization for different application scenarios.

[0068] In addition, to comprehensively evaluate the technical effectiveness of the present invention, a systematic experimental plan was designed, including dataset construction, comparison algorithm selection, evaluation metric setting, and result analysis. Regarding the dataset, 3,000 target tracks with four typical motion patterns (constant speed, constant acceleration, coordinated turning, and accelerated turning) were simulated within a 10km radius radar measurement area. A balanced dataset of 48,000 samples was constructed by varying interruption times and interception strategies to ensure sufficient training and testing. The currently top-performing TSADCNN method was selected as the primary benchmark for the comparison algorithm. An ablation experiment (removing the assisted attention module) was also conducted to verify the contribution of the assisted attention module. The evaluation metric used was the track correct association rate, which is the ratio of correctly associated targets to the total number of targets. This is the most direct and effective performance metric in the track association problem. Experimental results demonstrate that the present invention performs well in all test scenarios. In the interruption time sensitivity test, 25 target tracks were randomly generated. As the interruption time increased from 4 seconds to 20 seconds, the present invention maintained a significant advantage. In particular, the accuracy remained at 96.4% for a 20-second interruption, nearly 20 percentage points higher than that of TSADCNN. This result verifies the effectiveness of the temporal feature extraction and interaction mechanism designed by the present invention for long interruption scenarios. In the multi-target scenario test, as the number of targets increases from 10 to 50, the performance of the present invention decreases the least, and still maintains an accuracy of 95.2% at 50 targets, while TSADCNN has dropped to 76.6%. This proves that the feature representation designed by the present invention has better discriminability and can effectively deal with track confusion problems in dense target scenarios. The results of the ablation experiment show that the assisted attention module brings an average performance improvement of 1.42% in different scenarios, and the improvement is more obvious in the most challenging scenarios (long interruptions, multiple targets), indicating that the module is particularly important for handling complex situations. In the new track length change test, the performance of the present application and the ablation experiment can maintain a high level, and the accuracy of the TSADCNN method is 11.6% lower than that of the present invention on average.

[0069] Combining all the experimental results, it can be concluded that the present application has achieved significant performance improvement on the key issue of interrupted track association through innovative network architecture and feature interaction mechanism, especially in the most challenging long interruption and multi-target scenarios, the advantages are more obvious. This enables it to better meet the demand for stable tracking of targets with high maneuverability and large changes in radar cross-section, which leads to interruption of batches, and has important military application value. Therefore, the present invention can be widely used in various types of air defense radar systems, and is particularly suitable for dealing with tracking interruption problems of targets with high maneuverability and long-term interruption. During actual deployment, the trained model can be integrated into the radar signal processing chain, and the newly detected track segments can be processed in real time and correlated and matched with the historical track database. The radar system can adjust the correlation threshold according to actual needs to balance the ratio of missed association and false association.

[0070] Step 1023 : In response to determining that the maximum track association probability value in the track association probability value sequence is less than the preset association probability, the accumulated interruption duration of the historical track is recorded.

[0071] In some embodiments, the execution entity may record the cumulative interruption duration of the historical track in response to determining that the maximum track association probability value in the track association probability value sequence is less than a preset association probability. The cumulative interruption duration is the duration from the interruption moment to the current moment. If the maximum track association probability value in the track association probability value sequence is less than the preset association probability, this may indicate that there is no new track associated with the historical track.

[0072] Step 1024 : In response to determining that the accumulated interruption duration is less than the preset interruption duration threshold, the new track group is updated, and the track association step is performed again.

[0073] In some embodiments, the execution entity may, in response to determining that the cumulative interruption duration is less than a preset interruption duration threshold, update the new track group and re-perform the track association step. The fact that the cumulative interruption duration is less than the preset interruption duration threshold (the batch interruption time threshold) may indicate that the cumulative interruption duration is within the batch interruption time threshold, and the currently acquired new tracks are not associated, thereby acquiring a new track to continue the association step.

[0074] Optionally, in response to determining that the maximum track association probability value in the above track association probability value sequence is greater than or equal to the above preset association probability, it is determined that the track association between the new track corresponding to the above maximum track association probability value and the above historical track is completed.

[0075] Optionally, in response to determining that the accumulated interruption duration is greater than or equal to the preset interruption duration threshold, the track association step is terminated, and the track association of the historical track is determined to have failed. The accumulated interruption duration being less than the preset interruption duration threshold may indicate that no new track associated with the historical track was found within the interruption time threshold.

[0076] The technical advantages of the present invention are mainly reflected in the following aspects: First, a breakthrough has been achieved in performance indicators, and the accuracy of interrupted track association has been increased to more than 95%, which is significantly better than the existing best method. This progress is of direct significance for improving the efficiency of air traffic control decision-making. Secondly, an innovative assisted attention mechanism is proposed to guide the feature learning of new tracks through old tracks (i.e. historical tracks), which solves the problem of interaction between features of sequences of unequal lengths and provides a new solution for similar time series association problems. Third, the network architecture design fully considers the actual application needs, the asymmetric structure adapts to the difference in length of old and new tracks, and the modular design facilitates adjustment and optimization for different scenarios. Fourth, the algorithm is highly robust and shows stable performance for changing factors such as interruption time, number of targets and length of new tracks, and is suitable for complex actual combat environments. Fifth, the transition from experience-driven to data-driven has been achieved, which has greatly reduced the need for manual parameter adjustment and improved the automation level and deployment efficiency of the system. Compared with the existing technology, the present invention has three significant features: first, the feature interaction is more complete, and the deep interaction of new and old track features is achieved by assisting the attention mechanism, rather than a simple loss function connection; second, the spatiotemporal feature extraction is more comprehensive, taking into account both spatial motion patterns and temporal evolution laws; third, it has stronger practical applicability, and is specifically designed for unequal length track segments commonly seen in air defense operations, solving the limitations of symmetric networks in this scenario.

[0077] Further references Figure 6 As an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of an interrupted track association device for comparing track motion characteristics. These device embodiments are similar to Figure 1 Corresponding to the method embodiments shown, the interrupted track association device for comparing track motion characteristics can be specifically applied to various electronic devices.

[0078] like Figure 6As shown, the interrupted track association device 600 for comparing track motion characteristics in some embodiments includes: an acquisition unit 601 and a track association unit 602 . Among them, the acquisition unit 601 is configured to acquire a new track group in response to detecting a historical track interruption; the track association unit 602 is configured to perform the following track association steps for the above-mentioned historical tracks and the above-mentioned new track group: select new tracks that meet preset screening conditions with the above-mentioned historical tracks from the above-mentioned new track group to obtain a target track group; input the above-mentioned target track group and the above-mentioned historical tracks into a preset track association neural network to generate a track association probability value sequence, wherein the above-mentioned track association neural network is used to extract motion features of new tracks and historical tracks for track association, each track association probability value corresponds to a target track, and the above-mentioned track association neural network includes: a preprocessing module, a track feature extraction module, an auxiliary attention module and a classifier module; in response to determining that the maximum track association probability value in the above-mentioned track association probability value sequence is less than the preset association probability, record the cumulative interruption time of the above-mentioned historical tracks; in response to determining that the above-mentioned cumulative interruption time is less than the preset interruption time threshold, update the above-mentioned new track group, and perform the above-mentioned track association step again.

[0079] It can be understood that the units recorded in the interrupted track association device 600 for comparing track motion characteristics are the same as those in the reference Figure 1 Therefore, the operations, features and beneficial effects described above for the method are also applicable to the interrupted track association device 600 for comparing track motion characteristics and the units included therein, and will not be described in detail here.

[0080] Reference below Figure 7 , which shows a schematic structural diagram of an electronic device (such as a radar device) suitable for implementing some embodiments of the present disclosure. Figure 7 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present disclosure. Figure 7 As shown, the computer device includes a processor, a memory and a network interface connected via a system bus, wherein the memory may include a non-volatile storage medium and an internal memory. The non-volatile storage medium may store an operating system and a computer program. The computer program includes program instructions, which, when executed, may enable the processor to execute any of the above methods. The processor is used to provide computing and control capabilities to support the operation of the entire computer device. The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium, which, when executed by the processor, may enable the processor to execute any of the above methods. The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 7The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present disclosure, and does not constitute a limitation on the computer device to which the scheme of the present disclosure is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0081] It should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0082] In one embodiment, the processor is configured to execute a computer program stored in a memory to implement the following steps: in response to detecting a historical track interruption, obtaining a new track group; performing the following track association step for the historical track and the new track group: selecting new tracks from the new track group that meet preset screening conditions with the historical track to obtain a target track group; inputting the target track group and the historical track into a preset track association neural network to generate a track association probability value sequence, wherein the track association neural network is configured to extract motion features of the new track and the historical track for track association, each track association probability value corresponding to a target track, and the track association neural network includes: a preprocessing module, a track feature extraction module, an auxiliary attention module, and a classifier module; in response to determining that the maximum track association probability value in the track association probability value sequence is less than a preset association probability, recording the cumulative interruption duration of the historical track; in response to determining that the cumulative interruption duration is less than a preset interruption duration threshold, updating the new track group, and performing the track association step again.

[0083] An embodiment of the present disclosure further provides a computer-readable storage medium, on which a computer program is stored. The computer program includes program instructions. The method implemented when the program instructions are executed can refer to the various embodiments of the method described above in the present disclosure.

[0084] The computer-readable storage medium may be an internal storage unit of the computer device described in the aforementioned embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a SmartMedia Card (SMC), a Secure Digital (SD) card, a flash memory card, etc., provided on the computer device.

[0085] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or system comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or system. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or system comprising the element.

[0086] The above description is only an illustration of some preferred embodiments of the present disclosure and the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with (but not limited to) technical features with similar functions disclosed in the embodiments of the present disclosure.

Claims

1. A method for associating interrupted tracks by comparing track motion characteristics, characterized in that: include: In response to detecting a historical track interruption, acquiring a new track set; For the historical track and the new track group, perform the following track association steps: Selecting new tracks that meet a preset screening condition with the historical tracks from the new track group to obtain a target track group; Inputting the target track group and the historical tracks into a preset track association neural network to generate a track association probability value sequence, wherein the track association neural network is used to extract motion features of the new track and the historical tracks for track association, each track association probability value corresponds to a target track, and the track association neural network includes: a preprocessing module, a track feature extraction module, an auxiliary attention module, and a classifier module; In response to determining that the maximum track association probability value in the track association probability value sequence is less than a preset association probability, recording the accumulated interruption duration of the historical track; In response to determining that the accumulated interruption duration is less than a preset interruption duration threshold, the new track group is updated, and the track associating step is performed again.

2. The method according to claim 1, characterized in that The method further comprises: In response to determining that the maximum track association probability value in the track association probability value sequence is greater than or equal to the preset association probability, it is determined that track association between the new track corresponding to the maximum track association probability value and the historical track is completed.

3. The method according to claim 1, characterized in that The method further comprises: In response to determining that the accumulated interruption duration is greater than or equal to the preset interruption duration threshold, the track association step is terminated, and it is determined that the track association of the historical track has failed.

4. The method according to claim 1, wherein The step of selecting a new track from the new track group that satisfies a preset screening condition with the historical track to obtain a target track group includes: Determining the interruption time point, interruption end coordinates and corresponding target interruption speed value of the historical track, and determining the starting point coordinates corresponding to each new track in the new track group; determining, based on the interruption time point, the interruption end coordinates, and the target interruption speed value, whether each new track in the new track group satisfies a preset screening condition, wherein the screening condition is that a distance value between the starting coordinates of the new track and the interruption end coordinates is less than or equal to a track distance, wherein the track distance is generated by the interruption time point, the target interruption speed value, and a preset speed difference; The new track that meets the preset screening condition is determined as the target track to obtain a target track group.

5. The method according to claim 1, characterized in that The step of inputting the target track group and the historical tracks into a preset track association neural network to generate a track association probability value sequence includes: For each target track in the target track group, perform the following steps: Inputting the target track and the historical track into a preprocessing module in the track association neural network for encoding processing to obtain historical route encoding features and new route encoding features; Extracting features of the historical route code and the new route code through the auxiliary attention module and the track feature extraction module to obtain a historical route feature vector and a new route feature vector; The historical route feature vector and the new route feature vector are classified by the classifier module to generate a track association probability value.

6. The method according to claim 5, characterized in that The track feature extraction module is composed of multiple identical layers stacked together, each layer containing a multi-head self-attention sublayer and a feedforward neural network sublayer. Each layer is equipped with residual connections and normalization layers. The multi-head self-attention captures sequence relationships from different subspaces by running multiple attention heads in parallel. The feedforward neural network provides nonlinear transformation capabilities. The track feature extraction module flattens each historical route feature vector and new route feature vector respectively, and performs dimensionality reduction processing through a multi-layer perceptron to obtain the historical route feature vector and the new route feature vector.

7. A device for associating interrupted tracks by comparing track motion characteristics, comprising: an acquisition unit configured to acquire a new track group in response to detecting a historical track interruption; The track association unit is configured to perform the following track association steps for the historical track and the new track group: Selecting new tracks that meet a preset screening condition with the historical tracks from the new track group to obtain a target track group; Inputting the target track group and the historical tracks into a preset track association neural network to generate a track association probability value sequence, wherein the track association neural network is used to extract motion features of the new track and the historical tracks for track association, each track association probability value corresponds to a target track, and the track association neural network includes: a preprocessing module, a track feature extraction module, an auxiliary attention module, and a classifier module; In response to determining that the maximum track association probability value in the track association probability value sequence is less than a preset association probability, recording the accumulated interruption duration of the historical track; In response to determining that the accumulated interruption duration is less than a preset interruption duration threshold, the new track group is updated, and the track associating step is performed again.

8. An electronic device comprising: one or more processors; A storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

9. A computer-readable medium having a computer program stored thereon, wherein: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images