A method, device and electronic equipment for predicting a break in a track

Abstract

The application discloses a kind of prediction methods, device and electronic equipment of interrupting track, it is related to track identification technical field, the method includes to the original track data of acquisition is preprocessed, secondly, preprocessed track data is distinguished and marked to obtain first track data and second track data, then first track data and second track data are calculated to the causal value of first track data and second track data, according to causal intensity maximum value, the track segment of first track data and the track segment of second track data are associated and paired to determine out prediction track, by calculating causal value, the track segment of first track data and the track segment of second track data are associated and paired using causal intensity maximum value, to improve the robustness and generalization ability of track interruption association.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of track recognition, in particular to a method and device for predicting interrupted tracks and electronic equipment. BACKGROUND

[0002] With the continuous expansion of global marine trade, the widespread use of equipment such as the automatic identification system (AIS), shore-based radar, and optical / electronic infrared sensors has led to an exponential increase in the amount of track data. However, due to the complex marine environment (such as severe weather, electromagnetic interference, and sensor blind spots) and differences in equipment sampling strategies, track data often exhibits discontinuity in time or space, i.e., the "track interruption" phenomenon. Traditional correlation methods mainly rely on simplified motion models such as uniform speed and uniform acceleration, and use Kalman filtering or interactive multiple models (IMM) to predict backward from old tracks and backtrack forward from new tracks. Correlation is achieved by matching based on distance, speed, or heading thresholds. Such methods can still be used in open sea or regular route scenarios, but in near-shore dense traffic areas, maneuvering avoidance scenarios, or non-cooperative target (AIS off) conditions, the correlation robustness significantly decreases due to model mismatch, noise sensitivity, and threshold rigidity.

[0003] In addition, due to the limited size and single scene of existing public data sets, and the lack of cross-sensor and cross-sea generalization evaluation benchmarks, the practical bottleneck is further exacerbated. Therefore, it is urgent to build a hybrid framework that combines physical constraints and data-driven advantages to improve the robustness and generalization ability of track interruption correlation while ensuring real-time performance and interpretability. SUMMARY

[0004] In view of the above, the embodiments of the present application provide a method and device for predicting interrupted tracks and electronic equipment to solve the problem of how to improve the robustness and generalization ability of track interruption correlation.

[0005] According to a first aspect, the embodiments of the present application provide a method for predicting interrupted tracks, comprising:

[0006] Obtaining original track data, preprocessing the original track data, and outputting preprocessed track data, wherein the preprocessed track data includes a track segment before an interrupted track and a track segment after the interrupted track;

[0007] Differentiating and marking the preprocessed track data, marking the track segment before the interrupted track as first track data, marking the track segment after the interrupted track as second track data, calculating the causal value of the first track data and the second track data, and determining the causal strength between the track segments in the first track data and the track segments in the second track data;

[0008] The track segment of the first track data and the track segment of the second track data corresponding to the maximum causal strength are associated and paired, and a predicted track is output.

[0009] The method for predicting interrupted track provided in the application first pre-processes the obtained original track data, secondly distinguishes and marks the pre-processed track data to obtain first track data and second track data, then calculates the causal value of the first track data and the second track data by using the first track data and the second track data, and determines the predicted track by associating and pairing the track segment of the first track data and the track segment of the second track data according to the maximum causal strength. By calculating the causal value and associating and pairing the track segment of the first track data and the track segment of the second track data according to the maximum causal strength, the robustness and generalization ability of the interrupted track association are improved.

[0010] According to a second aspect, an embodiment of the application provides a device for predicting interrupted track, which comprises:

[0011] An acquisition module is configured to acquire original track data, pre-process the original track data, and output pre-processed track data, wherein the pre-processed track data comprises a track segment before an interrupted track is generated and a track segment after the interrupted track is generated.

[0012] A marking module is configured to distinguish and mark the pre-processed track data, mark the track segment before the interrupted track as first track data, mark the track segment after the interrupted track as second track data, calculate the causal value of the first track data and the second track data, and determine the causal strength between the track segment in the first track data and the track segment in the second track data.

[0013] An output module is configured to associate and pair the track segment of the first track data and the track segment of the second track data according to the maximum causal strength principle, and output a predicted track.

[0014] The device for predicting interrupted track provided in the application pre-processes the original track data acquired by the acquisition module, then the marking module distinguishes and marks the pre-processed track data to obtain first track data and second track data, then the output module calculates the causal value of the first track data and the second track data, and determines the predicted track by associating and pairing the track segment of the first track data and the track segment of the second track data according to the maximum causal strength, so as to improve the robustness and generalization ability of the interrupted track association by using the causal relationship.

[0015] According to a third aspect, an electronic device is provided, comprising: a memory and a processor, which are connected with each other in communication, the memory stores computer instructions, and the processor executes the computer instructions to perform the track interruption prediction method in the first aspect or any one of the implementation forms of the first aspect. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort based on these drawings.

[0017] Figure 1 A simulation schematic diagram of the track interruption re-association problem in the prior art.

[0018] Figure 2 A flowchart of the track interruption prediction method provided by the embodiments of the present application.

[0019] Figure 3 Simulation verification results of the track interruption prediction method provided by the embodiments of the present application in a simple scenario.

[0020] Figure 4 Simulation verification results of the track interruption prediction method provided by the embodiments of the present application in a complex scenario.

[0021] Figure 5 Track association results of the embodiments of the present application in a real scenario AIS (Automatic Identification System) data set.

[0022] Figure 6 A structural schematic diagram of the track interruption prediction device provided by the embodiments of the present application.

[0023] Figure 7 A structural schematic diagram of the electronic device provided by the embodiments of the present application.

[0024] REFERENCE NUMERALS

[0025] 10 - acquisition module; 11 - marking module; 12 - output module; 21 - processor; 20 - memory. DETAILED DESCRIPTION

[0026] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0027] It should be noted that, as Figure 1 shown, it is a schematic diagram of the track interruption reassociation problem in the prior art. The so-called track interruption problem refers to judging the homology of tracks by using track information (including but not limited to speed, heading, etc.) provided by multiple sensors (at least two) through a suitable method. Figure 1 In the figure, the red and blue line segments represent four tracks measured by sensors, in which two old tracks have the same time range, two new tracks have the same time range, and the part between the old track and the new track is a part of missing data. In the real track, the two red segment tracks are homologous, and the two blue line segment tracks are homologous. The track interruption reassociation problem is to select a suitable method to correspond and associate the observed tracks in the case of missing data in the track interruption period. Such interruption hinders accurate estimation of the target motion state and introduces statistical errors (discrete track segments are incorrectly identified as multiple different targets) in target number assessment, thereby damaging the reliability of data fusion and battlefield situation awareness analysis, and even possibly further leading to decision-making errors of the command level. It is a common and key challenge in multi-source data fusion.

[0028] The embodiments of the present application provide a track interruption prediction method, device and electronic equipment. As Figure 2 shown, it is a flowchart of the track interruption prediction method provided by the embodiments of the present application. The track interruption prediction method provided by the embodiments of the present application is applied to an intelligent terminal or a control system (such as a PC device), and the track segment of the first track data and the track segment of the second track data are associated and paired by using the maximum causal strength value through calculating the causal value, so as to improve the robustness and generalization ability of track interruption association.

[0029] Continuing to refer to Figure 1 , the track interruption prediction method provided by the embodiments of the present application includes the following steps:

[0030] S10, obtaining original track data, pre-processing the original track data, and outputting pre-processed track data. The pre-processed track data includes a track segment before the interrupted track and a track segment after the interrupted track.

[0031] In this embodiment, the original track data can be determined by the user according to the analysis requirements; when the original data is used for midcourse track prediction, the recognition algorithm used may misjudge the data "lost" as "disappeared" or "reborn", and pre-processing the original track data can reduce or avoid the occurrence of misjudgment, in addition, pre-processing the original track can also reduce or avoid data errors caused by sensor measurement, data collection deviation, and collection time deviation. For example: solving the instantaneous pseudo-causal deviation caused by the coordinate instantaneous shift and height mutation of radar data; or solving the large difference in data collected on different sensors due to different times and different environments (1s vs 5s).

[0032] S11, the pre-processed track data is distinguished and marked, the track segment before the interrupted track is marked as first track data, and the track segment after the interrupted track is marked as second track data, the causal value of the first track data and the second track data is calculated, and the causal strength between the track segments in the first track data and the track segments in the second track data is determined.

[0033] In this embodiment, the pre-processed track data is distinguished and marked, which can realize the compression of computing power cost, and by taking the interrupted track as a segmentation point, the causal information can be directly recognized in the subsequent algorithm, avoiding the pseudo-correlation across the interruption as causal information. In addition, by calculating the causal value and the causal strength, not only can it be used to guide track prediction, but also can be used to exclude external interference sources, such as if the causal strength is significantly higher than the historical baseline, it can be determined that this interruption is not pure signal loss, but may be electronic interference, weather mutation, etc.

[0034] S12, the track segments of the first track data and the second track data corresponding to the maximum causal strength are associated and paired, and the predicted track is output.

[0035] In this embodiment, the track segments of the first track data and the second track data are associated and paired using the maximum causal strength, which realizes matching with physical causality instead of geometric distance, maximally reduces misassociation and reduces track fragments, to obtain a continuous and reliable predicted track as the final output.

[0036] Using the maximum causal strength for track prediction can reduce the misassociation rate, and using the causal strength can also take into account the implicit dynamic inertia (such as the acceleration of the ship and the turning law) into the matching index, thereby retaining more real motion characteristics; when predicting with the aid of the maximum causal strength, after a track is interrupted by a sensor for multiple times and for a short time, the N fragments can be automatically combined into one continuous track by pairing according to the maximum causal strength, thereby reducing track fragments and relieving data pressure.

[0037] The method for predicting interrupted track provided in the embodiments of the present application first pre-processes the obtained original track data, secondly distinguishes and marks the pre-processed track data to obtain first track data and second track data, and then calculates the causal values of the first track data and the second track data by using the first track data and the second track data, and determines the predicted track by associating and pairing the track segments of the first track data and the track segments of the second track data according to the maximum causal strength. By calculating the causal values and associating and pairing the track segments of the first track data and the track segments of the second track data according to the maximum causal strength, the robustness and generalization ability of the track interruption association are improved.

[0038] The method for predicting interrupted track provided in the present application further comprises the following steps in step S10:

[0039] S101, classifying the original track data based on the time range, the geographical space constraint, the ship type and the navigation state, and outputting the classified track data.

[0040] S102, constructing track segments of the classified track data by using the track timestamps, and outputting the pre-processed track data.

[0041] In the embodiments, the time range, the geographical space constraint and the ship type can be defined by the user, and the navigation state is collected by the sensor and uploaded to the terminal device, and the user extracts according to the demand. Taking the A sea area as an example, the geographical constraint can be set as 60-120°E of east longitude and 0-30°N of north latitude. The ship type can be merchant ship or warship, and the navigation state can be always stationary and berthing. In the embodiments, the data classification and track segment construction are performed to improve the data quality and reduce the pressure of algorithm power usage.

[0042] Before step S102 is performed, the following steps need to be performed:

[0043] S1001, performing data cleaning on the classified track data, and outputting the classified track data after data cleaning.

[0044] S1002, constructing track segments of the classified track data after data cleaning by using the track timestamps, and outputting the pre-processed track data.

[0045] In the embodiments, the data cleaning on the classified track data can be pre-setting the threshold of the time range, the geographical threshold of the geographical space constraint, the ship type limitation and the navigation state limitation or threshold. When the classified track data is greater than or less than the threshold range or meets the limitation condition, the corresponding classified track data is filtered out, so that the data cleaning is completed. By cleaning the classified track data, not only the algorithm power is released, but also the accuracy of track prediction is further improved, and the robustness of track interruption association is improved.

[0046] Optionally, when step S10 is performed, it further includes:

[0047] S103, aggregating or segmenting the classified track data to obtain track construction data.

[0048] S104, performing preset time length screening on the track construction data to obtain pretreated track data.

[0049] In the embodiment, the aggregation of the ship track can reduce the noise and redundant points in the ship track data, and the segmentation of the ship track can separate the mixed track and avoid the interference of different ships or segments, that is, the aggregation or segmentation of the ship track is to improve the data quality and facilitate the subsequent acquisition of more accurate predicted track; the preset time length screening on the track construction data is to eliminate too short or abnormal segments (such as instantaneous jump) and retain effective navigation data to enhance the reliability of subsequent analysis. In the embodiment, the execution of step S10 can not only enhance the track consistency, reduce the calculation complexity, adapt to the requirements of downstream tasks, and improve the robustness of the whole track prediction. For example, the ship track aggregation operation can smooth the track break caused by GPS drift or signal loss, so that the track is more consistent with the actual navigation logic, such as port entry and exit and route continuity; the pretreatment of the ship track can reduce the scale of the pretreated track data, such as removing invalid short-time track, to further reduce the calculation burden of subsequent machine learning or visualization algorithm; the preset time length screening can realize flexible adjustment (such as retaining the track of ≥30 minutes).

[0050] The prediction method of the interrupted track provided in the application, the execution of step S11 further includes:

[0051] S111, extracting N time points close to the end part of the first track from the first track data;

[0052] S112, extracting N time points close to the start part of the second track from the second track data.

[0053] and,

[0054] S113, calculating the causal value of the first track data and the second track data by using the Liang's information flow causal analysis method.

[0055] In the embodiment, the N time points close to the end part of the first track and the N time points close to the start part of the second track can also be understood as obtaining the time sequence of the first track and the sequence of the second track, and preferably, the N time points are continuous time.

[0056] In the embodiment, the beam information flow causal analysis method is used to calculate the causal values of the first track data and the second track data, which can be represented by the following formula:

[0057]

[0058] In the formula, the represents the causal value from the first track data to the second track data; and represents the parameter matrix to be estimated related to the track data, c kj represents the sample covariance matrix composed of the first track data and the second track data, c′ ij represents the sample covariance matrix between the sequences derived by the Euler forward difference of the first track data and the second track data.

[0059] In the embodiment, T A→B represents the causal relationship between the first track data and the second track data of the subsystem, if T A→B is 0, it means that there is no causal relationship between A and B, if T A→B is significantly different from 0, it means that there is a significant causal relationship between A and B.

[0060] It can be understood that the prediction method of the interrupted track provided in the embodiment can quantitatively obtain the causal relationship between the two time series (the first track data and the second track data), that is, the size of the information transmission / information flow between them, that is, if the output result is not zero, it means that there is a causal relationship between the two (the first track data and the second track data), if is equal to zero, it can be considered that there is no causal relationship between the two.

[0061] The prediction method of the interrupted track provided in the application further comprises the following steps:

[0062] S121, obtaining the maximum causal value of the first track data and the second track data.

[0063] S122, extracting the track segment of the first track data corresponding to the maximum causal value and the track segment of the second track data.

[0064] S123, determining the track segment of the first track data and the track segment of the second track data as homologous tracks.

[0065] S124, correlating and splicing the homologous tracks according to the time sequence, and outputting the predicted track.

[0066] The method for predicting interrupted track provided in the application realizes the extraction of candidate pairs of spatiotemporal constraints through data preprocessing; the causal strength between the candidate pairs is quantified by using a causal analysis method based on Liang's information flow; and the homologous tracks are obtained by using the maximum matching principle. The track segments of the first track data and the track segments of the second track data are defined as two subsystems (time series), each of which is composed of elements such as a directional angle, a speed, and a coordinate. The correlation prediction between the interrupted tracks is realized by calculating the causal strength between the subsystems.

[0067] As shown in Figure 3 , it is a process diagram for verifying the causal analysis method in some simple scenarios to determine the feasibility of the interrupted track correlation problem. A monomial quadratic function f = x ^2 Two sequences are generated to represent two tracks, and the same color represents homologous tracks. Then, the data in the middle period is artificially removed, and it is assumed that the removed data is the interrupted data (interrupted track). The data before the interruption is referred to as the old track (first track data), and the data after the interruption is referred to as the new track (second track data). Then, the correlation between the old track segments and the new track segments is determined by using the causal value analysis method mentioned above, as shown in Figure 3 , the left half of Figure 3 represents the data with interrupted tracks, and the right half is the result of the interrupted track prediction realized by the method provided, which can be understood as achieving 100% correlation accuracy (the dashed line represents the correlation result).

[0068] As shown in Figure 4 , the method provided in the application is extended to high-complexity (such as multiple cross-interruptions) application scenarios, simulation data (constructed by an exponential function) is constructed, and the middle period is artificially removed as an interruption scenario. The same color still represents homologous tracks, and the dashed line represents the correlation result. Referring to Figure 4 , it can be known that the correlation of the interrupted tracks can still be realized by using the causal analysis method mentioned above, wherein Figure 4 the left half of Figure 4 represents the data with interrupted tracks, and the right half is the result of the interrupted track prediction realized by the method provided, which can be understood as achieving 100% correlation accuracy (the dashed line represents the correlation result).

[0069] Specifically, the method for predicting interrupted track provided in the application is applied to the open-source real data set AIS (Automatic Identification System for Ships), and the steps of the method for predicting interrupted track provided above are described, and the specific steps are as follows:

[0070] Step 1 (S10): Preprocessing data, extracting track candidate pairs

[0071] By extracting the track segments in the sea area and time interval of interest, the track data is classified according to different ship types and basic conditions such as speed, and then the repeated and abnormal data is cleaned up. Subsequently, the track segments are arranged according to the chronological order of the track timestamps, and the track correlation pairs are constructed according to the chronological order of the timestamps and the data cleaning results.

[0072] Step 2 (S11): Calculate the causal strength between track segments according to the extracted track correlation pairs

[0073] After the track correlation pairs are constructed in step 1, any one group of track preselected pair segments is reselected and marked as an old track (i.e., the first track data), and the last L time points of the old track are intercepted and recorded as Y1, where Y1 is a time series containing coordinate information.

[0074] And select any one group of track segments marked as a new track (i.e., the second track data), and intercept the first L time points of the new track and record them as Y2, where Y2 is also a time series containing coordinate information.

[0075] Then, the causal value between the old track Y1 and the new track Y2 is calculated and stored in the array T, and the causal relationship between A and B can be obtained by the following formula:

[0076]

[0077] a ij in the formula represents the parameter matrix to be estimated, which can be obtained by maximum likelihood estimation, c kj represents the covariance matrix of the sample, c' ij represents the sample covariance matrix between the sequences derived by the Euler forward difference between the track segments.

[0078] Step 3 (S12): Based on the principle of maximizing causal strength, select the track sequence pair with the strongest causal relationship according to the results calculated in step 2, and mark it as homologous track.

[0079] Iterate through all track preselected pairs, update and replace Y2, calculate the causal value between Y1 and different Y2 and store it in array T, then find the maximum value in array T and the time series Y1 and Y2 corresponding to the maximum value, mark the track pair Y1 and Y2 as homologous track, and remove it from the candidate pairs of preprocessed data. Then, repeat step 2 to iterate through all track preselected pairs until the correlation of all track segments is completed, output the homologous track, and mark it as predicted track.

[0080] The method is verified on an open-source real dataset AIS (Automatic Identification System). Firstly, the dataset is preprocessed, classified according to different interruption times, and candidate pairs are extracted through reasonable space-time constraints (equivalent to preliminary data cleaning, cleaning unreasonable track segments completely on the space-time threshold), then the causal analysis method mentioned in the technical scheme is used to calculate the causal relationship between each candidate pair, and finally the associated track is obtained through the maximum matching principle. As shown in Figure 5 , a group of real scene association results are selected for display, that is, the same color line segment represents the same track, and the dashed circle marked is the association result obtained by the method provided by the application. As shown in Figure 5 , it can be seen that even in a real scene, the track association method provided by the application can still accurately associate the interrupted tracks.

[0081] The advantages of the interrupted track prediction method provided by the embodiment of the application are:

[0082] 1. In the application, the causal relationship of the track data is constructed, the prior assumption of the target motion model in the traditional method is avoided and solved by using the causal strength, that is, it is not necessary to know the motion model of the ship in the early stage, and only the time sequence of the position coordinates (or the speed field, the heading angle, etc.) can be used to judge whether the target ship is homologous, so as to improve the generalization ability and the association ability of complex scenes.

[0083] 2. Since the causal relationship is used for track association, the high calculation amount problem in the training and reasoning stage of the machine learning track association algorithm is overcome, the decision-making interpretability problem damaged by the black box nature of machine learning is avoided, the interrupted track prediction method is improved in robustness, generalization ability, calculation efficiency and interpretability, and more complex scenes are applied.

[0084] It should be understood that the size of the serial number of each step in the above embodiment does not mean the order of execution, and the execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the application.

[0085] Correspondingly, please refer to Figure 6 , the embodiment of the application provides an interrupted track prediction device, which comprises:

[0086] An acquisition module 10 is used to acquire original track data, pre-process the original track data, and output pre-processed track data, wherein the pre-processed track data comprises a track segment before an interrupted track is generated and a track segment after the interrupted track is generated, and the detailed content is described with reference to step S10.

[0087] The marking module 11 is configured to mark the preprocessed track data, mark a track segment before the generation of the interrupted track as first track data, mark a track segment after the generation of the interrupted track as second track data, calculate a causal value of the first track data and the second track data, and determine a causal strength between the track segment in the first track data and the track segment in the second track data. Details are described in step S11.

[0088] The output module 12 is configured to associate and pair the track segment of the first track data and the track segment of the second track data according to a causal strength maximization principle, and output a predicted track. Details are described in step S12.

[0089] The prediction device for the interrupted track provided in the embodiment of the application is configured to preprocess the original track data obtained by the obtaining module, mark the preprocessed track data by the marking module to obtain the first track data and the second track data, calculate the causal value of the first track data and the second track data by the output module, and associate and pair the track segment of the first track data and the track segment of the second track data according to the maximum causal strength to determine the predicted track, so as to improve the robustness and generalization ability of the track interruption association by using the causal relationship.

[0090] The embodiment of the application further provides an electronic device. Figure 7 As shown in the figure, the electronic device can include a processor 21 and a memory 20, wherein the processor 21 and the memory 20 can be connected through a bus or other means, Figure 7 For example, the connection through the bus.

[0091] The processor 21 can be a central processing unit (CPU). The processor 21 can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or a combination of the above.

[0092] The memory 20 is a non-transitory computer readable storage medium, which can be used to store non-transitory software programs, non-transitory computer executable programs and modules, such as the program instructions / modules of the prediction method for the interrupted track in the embodiment of the application (for example, Figure 6The acquisition module 10, the marking module 11, and the output module 12 shown in the figure. The processor 21 performs various functional applications and data processing of the processor 21 by running the non-transitory software programs, instructions, and modules stored in the memory 20, that is, implements the interrupted track prediction method in the above-mentioned method embodiments.

[0093] The memory 20 can include a program storage area and a data storage area, wherein the program storage area can store an operating system and at least one application required by a function; the data storage area can store data created by the processor 21, etc. In addition, the memory 20 can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some embodiments, the memory 20 can optionally include a memory remotely arranged with respect to the processor 21, which can be connected to the processor 21 through a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0094] The one or more modules are stored in the memory 20 and, when executed by the processor 21, perform the interrupted track prediction method as Figure 2 in the embodiments shown.

[0095] The above-mentioned electronic device can correspond to the specific details Figure 2 The above-mentioned electronic device can correspond to the specific details

[0096] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by computer program instruction related hardware, and the program can be stored in a computer readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD) or a solid state drive (SSD), etc. The storage medium can also include a combination of the above-mentioned types of memories.

[0097] Although the embodiments of the present application are described in conjunction with the accompanying drawings, various modifications and changes can be made by those skilled in the art without departing from the spirit and scope of the present application, and such modifications and changes fall within the scope defined by the appended claims.

Claims

1. A method for predicting interrupted flight paths, characterized in that, include: Acquire raw track data, preprocess the raw track data, and output preprocessed track data. The preprocessed track data includes track segments before track interruption and track segments after track interruption. The preprocessed track data is distinguished and labeled. Track segments before track interruption are labeled as first track data, and track segments after track interruption are labeled as second track data. The causal value of the first track data and the second track data is calculated to determine the causal strength between track segments in the first track data and track segments in the second track data. The causal value between the first track data and the second track data is expressed by the following formula: ； Among them, in the formula This represents the magnitude of the causal value from the first track data to the second track data; and Represented as the parameter matrix to be estimated, It is represented as the sample covariance matrix composed of the first track data and the second track data. It is represented as the sample covariance matrix between the sequences of the first track data and the second track data derived using Euler's forward aberration; Extract the track segments of the first track data and the second track data corresponding to the maximum causal strength, associate and pair them, and output the predicted track.

2. The method for predicting interrupted flight paths according to claim 1, characterized in that, The original track data is preprocessed to output preprocessed track data, including: The original track data is classified based on time range, geospatial constraints, vessel type, and navigation status, and classified track data is output. The classified track data is used to construct track segments using track timestamps, and the preprocessed track data is output.

3. The method for predicting interrupted flight paths according to claim 2, characterized in that, Before constructing track segments from the categorized track data using track timestamps and outputting the preprocessed track data, the process further includes: The classified track data is cleaned, and the cleaned classified track data is output. The cleaned and categorized track data is used to construct track segments, and the preprocessed track data is output.

4. The method for predicting interrupted flight paths according to claim 3 or 2, characterized in that, Using track timestamps, track segments are constructed from the categorized track data, and the preprocessed track data is output, including: The classified track data is aggregated or segmented to obtain track construction data; The preprocessed track data is obtained by filtering the track construction data for a preset duration.

5. The method for predicting interrupted flight paths according to claim 1, characterized in that, The step of distinguishing and marking the preprocessed track data, marking the track segment before the track interruption as the first track data, and marking the track segment after the track interruption as the second track data, includes: Extract N time points from the first trajectory data that are close to the end of the first trajectory; Extract N time points from the second trajectory data that are close to the beginning of the second trajectory.

6. The method for predicting interrupted flight paths according to claim 1, characterized in that, The step of calculating the causality value of the first track data and the second track data, and determining the causal strength between track segments in the first track data and track segments in the second track data, includes: The causal value of the first track data and the second track data is calculated using Liang's information flow causal analysis method.

7. The method for predicting interrupted flight paths according to claim 1, characterized in that, The step of extracting the maximum causal strength corresponding to the track segment of the first track data and the track segment of the second track data, associating and pairing them, and outputting the predicted track includes: Obtain the maximum causality value of the first track data and the second track data; Extract the track segments of the first track data and the second track data corresponding to the maximum causal value; The track segments of the first track data and the track segments of the second track data are identified as tracks from the same source. The corresponding trajectories are associated and spliced ​​according to the time sequence, and the predicted trajectories are output.

8. A device for predicting interrupted flight paths, characterized in that, include: The acquisition module is used to acquire raw track data, preprocess the raw track data, and output preprocessed track data. The preprocessed track data includes track segments before the generation of the interrupted track and track segments after the generation of the interrupted track. A labeling module is used to distinguish and label the preprocessed track data, labeling track segments before track interruption as first track data and track segments after track interruption as second track data. It calculates the causality value between the first and second track data to determine the causal strength between track segments in the first and second track data. The causality value between the first and second track data is expressed by the following formula: ； Among them, in the formula This represents the magnitude of the causal value from the first track data to the second track data; and Represented as the parameter matrix to be estimated, It is represented as the sample covariance matrix composed of the first track data and the second track data. It is represented as the sample covariance matrix between the sequences of the first track data and the second track data derived using Euler's forward aberration; The output module is used to extract the track segments of the first track data and the second track data corresponding to the maximum causal strength, associate and pair them, and output the predicted track.

9. An electronic device, characterized in that, include: A memory and a processor are communicatively connected, the memory storing computer instructions, and the processor executing the computer instructions to perform the interrupted trajectory prediction method according to any one of claims 1-7.

Images