A sea-air target target grouping system

Abstract

The application discloses a sea-air target target grouping system, which mainly realizes grouping analysis and communication relationship analysis of sea-air targets in a sea-air environment. The application comprises an offline data analysis system and an online data analysis system. The same type targets can be grouped, and the analysis of different type target composite groups can also be performed. According to the type of data to be analyzed, the corresponding analysis system in the grouping system is selected, and the data is calculated through a grouping analysis model, so that the grouping result is obtained. Moreover, the result obtained by the offline data analysis system can also be analyzed and calculated through a communication relationship analysis model, so that the internal communication relationship of a target group is obtained.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of battlefield target grouping, in particular to a target grouping system for sea and air targets. The present application can be widely applied to complex sea and air battlefield environments for target grouping analysis of sea and air targets. BACKGROUND

[0002] Since ancient times, the action of the army has been carried out in units of combat clusters, and the modern military environment emphasizes this point. And due to the development of technology and the increasing advancement of combat facilities, a combat cluster contains more types of combat facilities belonging to different branches. This point is particularly evident in the sea battlefield environment, where sea and air combat facilities appear in coordination, and as a cluster to carry out activities has become the basic grouping situation in the modern sea battlefield environment. Therefore, how to effectively analyze the detected sea and air targets to obtain which targets and what kind of communication relationship as a combat cluster to carry out coordinated action not only can help predict the opponent's action intention and analyze the opponent's threat situation, but also can be used as a basis for formulating corresponding countermeasures. Now the target grouping prediction analysis is often based on the experience of front-line personnel, which not only requires the ability of relevant professionals, but also requires relevant personnel to work harder in the case of large and complex data, which has a certain negative impact on time efficiency and personnel use efficiency. Therefore, an automatic and efficient analysis model can effectively improve the work efficiency of manual target grouping work.

[0003] The existing patent CN111783020A discloses a target grouping method and system based on track data points, which performs DBSCAN density clustering on battlefield entity target track points, and the clustering radius is calculated according to a multi-dimensional Euclidean distance formula. This grouping method cannot well group different types of targets, and does not well handle the influence of the time dimension on grouping.

[0004] The existing patent CN110781963A discloses an air target grouping analysis method using K-means clustering, which combines the maximum and minimum distance algorithm with the K-means clustering algorithm, uses the maximum and minimum distance algorithm to select the initial clustering center for all targets, generates the number of target groups that are expected to be generated, and finally uses K-means clustering for air target grouping. However, this patent can only group and analyze air targets, and cannot analyze sea target grouping and the composite group of sea targets and air targets.

[0005] And the above existing patents do not involve a method for target grouping of online real-time data. SUMMARY

[0006] The present application is to solve the deficiencies of the prior art patent, and in view of the current situation that the target group gradually converts from the same type of air target group or sea target group to the sea-air target composite group in the current sea-air battlefield environment, a target grouping system for sea-air targets is involved. The system can start from the time domain and the space domain according to different data source conditions, and can not only group the same type of targets, but also group different types of targets. At the same time, the grouping results after grouping can also mine the internal communication relationship.

[0007] The system is composed of the following parts:

[0008] The offline data analysis system includes a grouping analysis module and a communication analysis module, which analyzes the grouping of a large amount of offline data, mines the target grouping relationship contained therein, and analyzes the internal communication relationship of the grouped cluster to mine the communication relationship contained therein.

[0009] The online data analysis system analyzes the real-time target grouping situation of the real-time data stream of sea-air targets, and gives the real-time grouping situation as the target data arrives.

[0010] Compared with the prior art, the present application has the following advantages

[0011] The present application can not only process the offline data accumulated in the past, but also analyze the real-time data stream. From the time-space domain and the target communication situation, the grouping result is more accurate. At the same time, different target type composite group grouping analysis can be carried out, which can cope with more complex grouping situations. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 The architecture of the system is shown in the figure;

[0013] Figure 2 The model composition and processing flow of the offline data analysis system are shown in the figure;

[0014] Figure 3 The processing flow of the target grouping analysis model is shown in the figure;

[0015] Figure 4 The architecture of the online data analysis system is shown in the figure. DETAILED DESCRIPTION

[0016] The present application will be described in further detail below with reference to the accompanying drawings.

[0017] Referring to Figure 1 The system architecture of the system includes a data access point, a grouping analysis system, an offline data analysis system and an online data analysis system, and a result display interface.

[0018] Referring to Figure 2The offline data analysis system is composed of a target group analysis model and a target communication analysis model. The offline data analysis system processing flow is as follows:

[0019] a) After obtaining data from the data access point, data preprocessing is performed, information content required by the target group analysis model is extracted, and then the information content is processed into a data form conforming to the data format requirements of the target group analysis model;

[0020] b) The data is analyzed by the target group relationship analysis model to obtain target group situations of the same type of target and composite group situations of different types of target;

[0021] c) The results obtained by using the target group relationship analysis model are processed, and for each cluster result, all communication modes of the units in the cluster are sorted according to the communication time to obtain the cluster communication data set;

[0022] d) The target communication relationship analysis model is used to analyze the cluster communication data set of each cluster to obtain the communication situation of each cluster.

[0023] The target group relationship analysis model in the offline data analysis system performs target group analysis calculation in the flow shown in Figure 3 , and the specific flow method is as follows:

[0024] a) First, the target data is read and preprocessed. The preprocessing method is to sort the data according to the first occurrence time of the data in a time interval (one day, several days, or one week, etc.) to generate an ordered sequence set;

[0025] b) The generated ordered sequence set is processed by the PrefixSpan algorithm to mine the frequent sequence set formed by the target platform;

[0026] c) The frequent sequence set mined in the previous step is pruned to retain non-repeating parent sets. The sequence elements are divided and extracted according to the target platform type to generate new sequences;

[0027] d) The several sequences of the same target type obtained in the previous step are subjected to trajectory clustering using the TraClus clustering algorithm. The parameters of the clustering algorithm are set according to the corresponding guide parameters of the target type of the target in the sequence to obtain the group situation of the same type of target and the representative trajectory of each group;

[0028] e) The same type of target group situation obtained in the previous step is analyzed and calculated using a different type of target group situation analysis algorithm to obtain the target group situation of the composite type target of different types of target, which is the final result.

[0029] The analysis algorithm of different types of target group mentioned in the above process e) is as follows:

[0030] a) First, use the aircraft group and ship group analysis method to obtain a part of the possible aircraft and ship mixed cluster, and add it to the result set;

[0031] b) Then use the aircraft isolated target and ship group analysis method and the ship isolated target and aircraft group analysis method respectively to obtain a part of the possible aircraft and ship mixed cluster, and add it to the result set;

[0032] c) Finally, remove the mixed group from the original aircraft group result and ship group result, and add the remaining group result to the result set;

[0033] d) Output the result set as the final analysis result of the target group situation analysis module.

[0034] The analysis method mentioned in the above process: for the grouped target, use its group representative trajectory, for the ungrouped isolated target, use its trajectory, first segment all trajectories according to their characteristics, then compare the similarity of trajectory segments of different types of targets, and use the Euclidean distance involving multiple dimensions as the measure of similarity. Then calculate the proportion of the number of similar trajectory segments of the corresponding group or isolated target and the number of all trajectory segments of other different types of target groups, and if the proportion meets the threshold, it is judged that the two belong to a cluster.

[0035] The target communication relationship analysis model in the offline data analysis system, the specific analysis process is as follows:

[0036] a) First, traverse the data set, when the traversal is completed, obtain all possible communication pairs of communication methods, and mark the target units that may be the highest level commander, and proceed to b);

[0037] b) Perform flag ship search, the search method is to check the communication method of all ship type targets, if a target of a certain communication method is much more than other communication methods, it is identified as a flagship, and added to the commander target set. If there is no possible highest level commander in a), use the flagship search method to find possible commander targets for other types of targets. If there is a possible highest level commander, proceed to c), otherwise proceed to d);

[0038] c) Begin a diffusion search centered on the highest-level command. Count its direct communication units and mark any non-command targets within them. Starting with the marked target, count and mark all non-command targets in its communication pairs. Continue searching for and marking other targets in the newly marked target's communication pairs. Repeat this search and marking process until all communication pairs are marked or no new markable objects can be found. At this point, it is considered that all possible communication scenarios at all communication levels have been found, and the algorithm terminates.

[0039] d) Assuming all units communicate at the same level, no further statistical search is performed on the communication pairs, and the algorithm ends.

[0040] Online data clustering analysis system such as Figure 4 As shown, when the data type of the data access point is real-time streaming data, the online data clustering analysis system is started to analyze the data through the online analysis model to obtain the real-time target clustering situation. The clustering analysis method used by the online analysis model is as follows:

[0041] a) Check the arrival data of the target and update the trajectory information. If the number of arrival data of the target meets the condition, then proceed to b).

[0042] If the current communication method for the arriving target exists, the target trajectory information table corresponding to that communication method is located. The arriving target information (including target name, time, and trajectory information) is filled into the table. The table is then searched forward, calculating the trajectory similarity for other targets that satisfy the time interval constraint. If other target trajectories satisfy the constraint exist, they are grouped into the same cluster as the arriving target. If the target was previously grouped into another cluster during arrival calculations, it is removed from that cluster. During the forward search, all trajectory information in the table that does not satisfy the time interval constraint is simultaneously cleared.

Claims

1. A target grouping system for air and sea targets, characterized in that: It consists of two parts: an offline data analysis system and an online data analysis system. An offline data analysis system, including a target grouping relationship analysis model and a target communication relationship analysis model; Based on previously accumulated offline data, and taking into account the spatiotemporal information contained in the data, we can extract the clustering of similar targets and the composite clustering of different types of targets. Furthermore, an intra-cluster connectivity analysis was conducted on the obtained clustering results to uncover the intra-cluster connectivity situation. The online data analysis system targets real-time data streams, taking into account the spatiotemporal information and communication conditions contained in the data, to mine the target clustering in real time and dynamically update the target clustering. The analysis and processing method for the target grouping relationship analysis model in the offline data analysis system is as follows: a) First, read the target data and preprocess it. The preprocessing method is to sort the data according to the first occurrence time of the data in a certain time interval to generate an ordered sequence set. b) The generated ordered sequence set is processed using the PrefixSpan algorithm to mine the frequent sequence set formed by the target platform; c) Prune the frequent sequence set mined in the previous step, retaining the non-repeating parent set; The sequence elements are then divided and extracted according to the target platform type to generate new sequences; d) For the several sequences of the same target type obtained in the previous step, use the TraClus clustering algorithm to perform trajectory clustering. The parameters of the clustering algorithm are set according to the corresponding guidance parameters of the target type in the sequence to obtain the grouping of the same type of targets and obtain the representative trajectory of each group. e) Analyze and calculate the target grouping situation of the same type of target obtained in the previous step using the target grouping situation analysis algorithm of different types of target to obtain the target grouping situation of composite type targets of different types of targets, which is the final result.

2. The air and sea target grouping system according to claim 1, characterized in that: The analysis and processing method for the target connectivity analysis model in the offline data analysis system is as follows: a) First, traverse the dataset. After the traversal is complete, obtain all possible communication pairs for all communication methods, and mark the target units that may be the highest-level command units, and then proceed with b). b) Perform flagship search. The search method is to check the communication methods of all ship type targets. If a target has a certain communication method far more than its other communication methods, it is identified as the flagship and added to the commander target set. If no target that may be the highest-level commander appears in the statistics in a), the flagship search method is used to find possible commander targets for other types of targets. If a possible highest-level commander is found in the end, proceed to c); otherwise, proceed to d). c) Starting with the highest-level command as the center, begin a diffusion search, count its direct communication units, and mark non-command targets. Starting from the marked target, count and mark all non-command targets in its communication pairs. Continue to search for and mark other targets in the newly marked target communication pairs. Repeat the above search and marking process until all communication pairs are marked or no new markable objects can be found. When all possible communication situations of all communication levels have been found, the algorithm ends. d) Assuming all units communicate at the same level, no further statistical search is performed on the communication pairs, and the algorithm ends.

3. The air and sea target grouping system according to claim 1, characterized in that: The cluster analysis method in the online data analysis system is as follows: a) Check the arrival data of the target and update the trajectory information. If the number of arrival data of the target meets the condition, then proceed to b). b) If the current communication method for the target exists, find the target trajectory information table corresponding to the communication method, fill the target information into the table, and then search the table forward. Calculate the trajectory similarity of the target information of other targets that meet the time interval constraint. If there are other target trajectories that meet the condition, group them together with the target. If the target was grouped into another group during the previous arrival calculation, delete it from the previous group. During the forward search, clear all trajectory information in the table that does not meet the time interval constraint.

Images