False track identification method and system based on multi-radar detection

Abstract

The invention discloses a false track identification method and system based on multi-radar detection, and relates to the technical field of radio navigation. The method comprises the following steps: deploying a data acquisition unit in a radar in a radar detection area, acquiring original trace point data of the radar in a preset monitoring period, transmitting the original trace point data of the radar to a data fusion center, and constructing a joint observation matrix of the radar detection area; then, steady-state observation relation analysis is carried out on the observation association graph sequence, a spatial steady-state feature library is dynamically updated, radar detection area track data generated by a data fusion center are collected, and the spatial steady-state feature library is combined to obtain a radar detection area track data fusion model; and finally, processing to obtain the type of the track of the radar detection area, and performing confidence feedback updating on the spatial steady-state feature library of the radar detection area, so that the track identification accuracy and reliability of multi-radar cooperative detection are improved.

Description

Technical Field

[0001] This invention relates to the field of radio navigation technology, specifically to a method and system for identifying false tracks based on multi-radar detection. Background Technology

[0002] With the increasing demand in fields such as air surveillance and low-altitude security, cooperative detection networks composed of multiple fixed radars have become a core technical means to achieve high-precision target tracking. The effective operation of cooperative detection networks depends on the accurate fusion and correlation of track data from different radars. Its core task is to improve the tracking accuracy and reliability of moving targets by comprehensively utilizing multi-view observation information through spatiotemporal alignment and data correlation. Traditional multi-radar data fusion methods are generally based on the principle of multi-source spatiotemporal consistency, which assumes that the observation results of multiple independent sensors on the same physical entity at the same time should be consistent in terms of spatial position and motion state, and track correlation decisions are made accordingly.

[0003] However, in real-world scenarios where multiple radars work together, especially in airspace with dense targets and complex maneuvers, the current mainstream fusion algorithms based on this principle have inherent limitations. The core of these algorithms is to find the optimal match for tracks from different radars within a pre-defined association decision. When multiple real target tracks intersect and run in parallel, or when a radar generates a series of trending false tracks due to interference such as multipath effects, the predicted positions of these track segments originating from different physical sources and independent of each other may coincide accidentally and continuously in space and time within the algorithm window.

[0004] At this point, based on the principle of observation consistency, the system will forcibly associate and splice these essentially unrelated fragments to generate a consensus track that may be discontinuous in a single radar view but appears to be continuous and complete in the fused global situation map. Such false tracks derived from the system's fusion logic are difficult to detect by conventional means because they conform to the surface characteristics of multi-source verification. Once generated, they will be continuously maintained, causing subsequent situation assessments and action guidance based on this erroneous information to lose their reliable basis, thereby affecting the effectiveness of network detection. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for identifying false flight paths based on multi-radar detection, which can effectively solve the problems mentioned in the background technology.

[0006] To achieve the above objectives, the first aspect of the present invention provides a method for identifying false flight paths based on multi-radar detection, comprising:

[0007] S1. Deploy a data acquisition unit on the radar in the radar detection area. Within a preset monitoring period, collect the original radar point data through the data acquisition unit and transmit the original radar point data to the data fusion center.

[0008] S2. The data fusion center receives the original radar point data, constructs a joint observation matrix of the radar detection area, and generates an observation correlation map sequence of the radar detection area arranged in time series.

[0009] S3. Perform steady-state observation relationship analysis on the observation correlation map sequence of the radar detection area to obtain the spatial steady-state feature points of the radar detection area. Update the feature points of the spatial steady-state feature library of the radar detection area based on the spatial steady-state feature points of the radar detection area.

[0010] S4. Collect radar detection area track data generated by the data fusion center, combine it with the spatial steady-state feature library of the radar detection area, perform logical anomaly detection and feature matching analysis on the radar detection area track, and obtain the risk level of the radar detection area track.

[0011] S5. Based on the risk level of the radar detection area track, determine the type of radar detection area track, and update the confidence level of the spatial steady-state feature database of the radar detection area.

[0012] Furthermore, the method for constructing the joint observation matrix of the radar detection area includes: after the data fusion center receives the original point data of the radar, it divides the radar detection area into a three-dimensional spatial grid of a preset size, and generates a corresponding observation vector for each radar in each monitoring cycle. The length of the observation vector is equal to the total number of three-dimensional spatial grid cells, and each vector element of the observation vector corresponds to a three-dimensional spatial grid cell.

[0013] If at least one original point of the corresponding radar falls into the currently processed three-dimensional spatial grid cell, the vector element corresponding to the three-dimensional spatial grid cell is marked as having an observation label; otherwise, it is marked as having no observation label. The observation vectors of all radars in the monitoring period are combined to form a joint observation matrix of the radar detection area in the corresponding monitoring period. Each row of the joint observation matrix corresponds to a radar according to a predefined radar sorting rule, and each column corresponds to a three-dimensional spatial grid cell according to a predefined spatial grid coding rule.

[0014] Furthermore, the method for generating a time-series sequence of observation correlation maps of radar detection areas includes: based on the joint observation matrix of the radar detection areas, calculating the correlation degree between the observation vectors of all different radar pairs in the joint observation matrix, constructing an observation correlation map, wherein the correlation degree is used to quantify the synchronization degree of two radars in spatial observation, and the radar pair is a combination of any two different radars in the joint observation matrix of the radar detection areas. The construction of the observation correlation map is performed for each monitoring cycle to generate a time-series sequence of observation correlation maps of radar detection areas.

[0015] Furthermore, the method for performing steady-state observation relationship analysis on the observation correlation map sequence of the radar detection area includes: extracting the correlation degree of each radar pair in each monitoring period from the observation correlation map sequence of the radar detection area arranged in time series, forming a correlation degree time series of the corresponding radar pair, and setting a correlation degree duration threshold, a correlation degree variance threshold, and a continuous period number threshold.

[0016] When a radar pair has a correlation time series that satisfies the following criteria, it is determined that there is a steady-state observation relationship between the radar pairs. The criteria are: within a continuous monitoring period of not less than the number of continuous periods threshold, the correlation degree of the radar pairs is higher than the correlation degree duration threshold, and the variance of the correlation degree of the radar pairs is lower than the correlation degree variance threshold within the continuous monitoring period.

[0017] Furthermore, the method for updating feature points in the spatial steady-state feature library of the radar detection area includes: analyzing radar pairs in the observation association sequence of the radar detection area that are determined to have a steady-state observation relationship, obtaining spatial steady-state feature points of the radar detection area, and recording the center coordinates, spatial distribution range, and corresponding radar pair identifier of the spatial steady-state feature points; determining whether the spatial steady-state feature points exist in the spatial steady-state feature library; if the spatial steady-state feature points do not exist in the spatial steady-state feature library, recording the feature labels of the spatial steady-state feature points, the feature labels of the spatial steady-state feature points including the center coordinates, spatial distribution range, corresponding radar pair identifier, and initial confidence value of the spatial steady-state feature points, and adding the feature labels of the spatial steady-state feature points to the spatial steady-state feature library of the radar detection area.

[0018] If the radar pair associated with the existing spatial steady-state feature point in the spatial steady-state feature library is still determined to have a steady-state observation relationship during the current monitoring period, then a confidence increase operation is performed, and the confidence value in the feature label of the corresponding spatial steady-state feature point is updated.

[0019] If the radar pair associated with the spatial steady-state feature point is no longer determined to have a steady-state observation relationship within the current monitoring period, a confidence reduction operation is performed, and the confidence value in the feature label of the corresponding spatial steady-state feature point is updated.

[0020] Furthermore, the method for collecting radar detection area track data generated by the data fusion center includes: the data fusion center receiving the original point track data of the radar, performing time-series correlation matching to obtain radar detection area track data, wherein the radar detection area track data includes radar detection area track, track identifier, and point track sequence arranged in chronological order that constitutes the radar detection area track.

[0021] Furthermore, the method for performing logical anomaly detection and feature matching analysis on the trajectory in the radar detection area includes:

[0022] Step a1: Analyze the radar detection area track data to obtain the key features of the radar detection area track. The key features of the radar detection area track include the radar composition characteristics and the space stationary core area of ​​the radar detection area track.

[0023] Step a2: Based on the radar detection area track data, perform logic anomaly detection on the radar detection area track, including timing anomaly analysis and motion anomaly analysis.

[0024] Temporal anomaly analysis and motion anomaly analysis are performed on the tracks in the radar detection area to obtain the temporal source anomaly determination results and motion anomaly determination results of the tracks in the radar detection area. The risk level of the tracks in the radar detection area is then obtained through processing.

[0025] The time-series source anomaly determination results include the presence of time-series source anomalies and the absence of time-series source anomalies; the motion anomaly determination results include the presence of motion anomalies and the absence of motion anomalies; and the risk level of the radar detection area track includes high-risk level and normal-risk level.

[0026] If the timing source anomaly determination result of the radar detection area track indicates the presence of a timing source anomaly, or the motion anomaly determination result of the radar detection area track indicates the presence of a motion anomaly, then the corresponding radar detection area track is assigned a high-risk level; otherwise, the corresponding radar detection area track is assigned a normal risk level.

[0027] Step a3: Perform feature matching analysis on the key features of the radar detection area track and the feature labels of each spatial steady-state feature point in the spatial steady-state feature library to obtain the source radar matching degree and spatial overlap degree between the radar detection area track and each spatial steady-state feature point. If the source radar matching degree between the radar detection area track and a spatial steady-state feature point is higher than the source radar matching degree threshold, and the corresponding spatial overlap degree is higher than the spatial overlap threshold, then the corresponding spatial steady-state feature point is recorded as the matching spatial steady-state feature point of the radar detection area track, and a high-risk level is assigned to the corresponding radar detection area track; otherwise, a normal risk level is assigned to the corresponding radar detection area track.

[0028] Furthermore, the method for determining the type of flight track in the radar detection area includes:

[0029] Step b1: For radar detection area tracks that are assigned a high-risk level due to the time-series source anomaly determination result or the motion anomaly determination result, the radar detection area track type is determined to be a dynamic anomaly type false track.

[0030] Step b2: For radar detection area tracks that are assigned a high-risk level due to feature matching analysis with the spatial steady-state feature library, the type of radar detection area track is determined to be a fixed false source type false track.

[0031] Step b3: If the radar detection area track simultaneously meets the determination conditions of steps b1 and b2, then the radar detection area track is determined to be a composite false track.

[0032] Step b4: Otherwise, determine that the type of the track in the radar detection area is a real track.

[0033] Furthermore, the false track identification method based on multi-radar detection also includes an initial construction method for the spatial steady-state feature library of the radar detection area, including: performing steady-state observation relationship analysis on the observation correlation map sequence of the radar detection area within the historical monitoring period to obtain the historical spatial steady-state feature points of the radar detection area, recording the feature labels of the historical spatial steady-state feature points, and constructing the spatial steady-state feature library of the radar detection area, which contains the feature labels of the historical spatial steady-state feature points of the radar detection area.

[0034] A second aspect of the present invention provides a false track identification system based on multi-radar detection, comprising:

[0035] The spot data acquisition module is used to deploy data acquisition units on radars in the radar detection area. Within a preset monitoring period, the data acquisition units collect the original spot data of the radar and transmit the original spot data of the radar to the data fusion center.

[0036] The correlation graph generation module is used to receive the original radar point data through the data fusion center, construct the joint observation matrix of the radar detection area, and generate the observation correlation graph sequence of the radar detection area arranged in time series.

[0037] The feature library update module is used to perform steady-state observation relationship analysis on the observation correlation map sequence of the radar detection area, obtain the spatial steady-state feature points of the radar detection area, and update the spatial steady-state feature library of the radar detection area based on the spatial steady-state feature points of the radar detection area.

[0038] The risk analysis module is used to collect radar detection area track data generated by the data fusion center, combine it with the spatial steady-state feature library of the radar detection area, perform logical anomaly detection and feature matching analysis on the radar detection area track, and obtain the risk level of the radar detection area track.

[0039] The judgment and analysis module is used to determine the type of radar detection area track based on the risk level of the radar detection area track, and to update the confidence value of the spatial steady-state feature library of the radar detection area.

[0040] The present invention has the following beneficial effects:

[0041] (1) By constructing a joint observation matrix and observation correlation graph sequence, the present invention transforms the analysis object into the collaborative observation relationship of the radar network itself. This method can systematically quantify and characterize the synchronization characteristics and temporal evolution of multiple radars in spatial coverage, thereby realizing the direct detection and positioning of abnormal modes caused by fixed reflectors, such as communication towers, which continuously generate highly synchronized false signals among multiple radars from within the system.

[0042] (2) In judging the risk level of a flight track, this invention analyzes the logic of the flight track itself from the perspective of time sequence anomalies and motion anomalies, and matches it with the spatial steady-state feature library. It can effectively identify dynamic logic anomalies caused by system misassociation and complex interference, as well as false targets generated by fixed reflectors in the environment. This improves the identification range and accuracy of false flight tracks. Compared with the existing technology, it can more comprehensively and accurately determine the authenticity of the flight track.

[0043] (3) The feedback update mechanism of the present invention gives the system the ability to continuously optimize itself. The resulting processing flow enhances the long-term robustness of the network detection. The spatial steady-state feature library will adjust the confidence and self-purify according to the final judgment result of the track, so that the system can accumulate experience, correct cognition and adapt to environmental changes. The system outputs track classification results with verified credibility to the upper-level decision module, so that it can perform differentiated processing based on clear property judgment, thereby improving the reliability and availability of the situation information output by the entire detection network. Attached Figure Description

[0044] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0045] Figure 2 This is a schematic diagram of the system module connections of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] Please see Figure 1 As shown, this embodiment of the invention provides a method for identifying false flight paths based on multi-radar detection, including:

[0048] S1. Deploy a data acquisition unit on the radar in the radar detection area. Within a preset monitoring period, collect the original radar point data through the data acquisition unit and transmit the original radar point data to the data fusion center.

[0049] Radar refers to a detection device that is fixedly deployed within the radar detection area and is capable of detecting aerial targets and generating raw point data. It is a basic node in the radar cooperative detection network.

[0050] The data acquisition unit refers to the software module deployed on each radar for real-time acquisition and structured processing of the raw point data output by the radar signal processor. The structured processing includes data encapsulation and timestamp normalization.

[0051] The radar's raw point data includes the radar's raw points within a preset monitoring period, the timestamp of the raw points, three-dimensional spatial coordinates, and source identifier. The radar's raw points refer to the single target observation report output by the signal processor of a single radar within one detection period, without fusion processing. It is the most basic and uncorrelated observation unit formed by the radar after detecting the reflected signals of potential targets in the air. The timestamp of the raw points refers to the precise absolute time when the raw points were detected by the radar. The three-dimensional spatial coordinates refer to the three-dimensional position information of the raw points in a unified geographic coordinate system obtained by transforming the coordinate system. The source identifier refers to the unique identification code of the radar that generated the raw points, such as the radar ID.

[0052] A data fusion center is a platform located in a radar cooperative detection network that receives raw radar spot data transmitted by data acquisition units deployed by various radars and performs fusion processing on this data.

[0053] S2. The data fusion center receives the original radar point data, constructs a joint observation matrix of the radar detection area, and generates an observation correlation map sequence of the radar detection area arranged in time series.

[0054] Specifically, the joint observation matrix of the radar detection area is constructed, and the specific process is as follows:

[0055] After receiving the raw radar data, the data fusion center divides the radar detection area into a three-dimensional spatial grid of preset size. In each monitoring cycle, it generates a corresponding observation vector for each radar. The length of the observation vector is equal to the total number of three-dimensional spatial grid cells. Each vector element of the observation vector corresponds to a three-dimensional spatial grid cell. By spatial gridding and generating observation vectors, the complex radar detection area is discretized into easily processed cells, enabling the observation information of each radar to be accurately represented in vector form, thus improving the accuracy and operability of data processing.

[0056] In a specific embodiment, assuming the radar detection area is divided into a unified grid of 32 units (4 east-west, 4 north-south, and 2 height), within a certain monitoring period, the observation vector of a radar is a sequence containing 32 elements, where each element corresponds to a specific grid unit. If the radar has one or more original point coordinates falling into a certain grid unit within this period, such as the 2nd east grid, the 3rd north grid, and the 1st height grid, then the vector element corresponding to that grid unit is marked as 1, indicating an observation marker. Conversely, if the radar has no points falling into a certain grid unit within this period, then the corresponding element is marked as 0, indicating no observation marker. The radar's observation of the entire airspace within this period is compressed and precisely represented as an observation vector of length 32, consisting of 0s and 1s.

[0057] It should be added that the preset size is a configurable system parameter. Based on the known performance parameters of the networked radar, such as range resolution, angle measurement accuracy, coverage area, and preset typical mission scenarios, such as target density and speed, those skilled in the art can use computer simulation to test the recognition rate of false tracks and system processing latency under different grid sizes. The optimized value of the performance inflection point is selected as the initial setting of the preset size. After the actual deployment of the system, flight tests are conducted using beacons or cooperative targets at known locations. Based on the actual data fusion effect and system resource usage, the initial setting of the preset size is fine-tuned and finally solidified to obtain the preset size. The method of determining key system parameters by combining computer simulation and actual measurement is a well-known and commonly used engineering optimization path in the field when deploying complex detection systems.

[0058] If at least one original point trace of a corresponding radar falls into the currently processed 3D spatial grid cell, the vector element corresponding to that 3D spatial grid cell is marked as having an observation label; otherwise, it is marked as having no observation label. This method of marking with different labels can effectively distinguish whether the radar has observed a point trace in each grid cell. The observation vectors of all radars within a certain monitoring period are combined to form a joint observation matrix of the radar detection area within that monitoring period. Each row of the joint observation matrix corresponds to a radar according to a predefined radar sorting rule, and each column corresponds to a 3D spatial grid cell according to a predefined spatial grid coding rule. The joint observation matrix integrates the spatial observation information of all radars within the same monitoring period, transforming the original point trace data into a matrix form that is easy to analyze, facilitating subsequent quantification of the observation relationships between radars, and thus generating a sequence of observation correlation maps of radar detection areas arranged in a time series.

[0059] In one specific embodiment, the predefined radar sorting rule is the source identifier of each radar in the radar cooperative detection network, such as radar ID: R001, R002, ..., which is sorted in ascending order according to its ID value. The predefined spatial grid coding rule is to divide the entire three-dimensional radar detection area into equally spaced grid cells along the east-west direction, the north-south direction, and the height direction, and to perform three-dimensional linear coding on all grid cells. The coding order is as follows: first, it increases from west to east along the east-west direction, then it increases from south to north along the north-south direction at the same east-west position, and finally it increases from low to high along the height direction at the same east-west position and the north-south position.

[0060] Specifically, the process of generating a sequence of observation correlation maps of radar detection areas arranged in time series is as follows:

[0061] Based on the joint observation matrix of the radar detection area, the correlation degree between the observation vectors of all different radar pairs in the joint observation matrix is ​​calculated to construct an observation correlation graph. This correlation degree is used to quantify the synchronization degree of the two radars in spatial observation, providing data support for judging the observation synchronization status of the radar cooperative detection network. The radar pair is a combination of any two different radars in the joint observation matrix of the radar detection area. Based on the calculated correlation degree between all radar pairs, an image construction algorithm is used to construct a weighted undirected graph with radars as nodes and correlation degree as edge weights. This weighted undirected graph is the observation correlation graph that represents the observation synchronization relationship of the radar cooperative detection network within the monitoring period. The observation correlation graph construction process is performed for each monitoring period to generate a time-series sequence of observation correlation graphs for the radar detection area. The time-series sequence of observation correlation graphs can reflect the changes in the observation synchronization relationship of the radar cooperative detection network over time, which helps to discover potential patterns and anomalies in the observation trajectory.

[0062] The correlation between the observation vectors of all different radar pairs is calculated as follows:

[0063] For any two different radars whose correlation is to be calculated, denoted as Radar A and Radar B, their observation vectors within the same monitoring period are extracted. The total number of units in the same spatial grid cell where all vector elements are marked with observation identifiers is counted, and this total number is recorded as the number of common observation units for the radar pair. The total number of vector elements marked with observation identifiers in Radar A's observation vector is obtained as the number of independent observation units for Radar A, and the total number of vector elements marked with observation identifiers in Radar B's observation vector is obtained as the number of independent observation units for Radar B. The number of common observation units is divided by the geometric mean of the number of independent observation units for Radar A and Radar B. The resulting ratio is the correlation between the observation vectors of Radar A and Radar B within the monitoring period. This correlation is used to quantitatively characterize the degree of synchronization between the two in spatial observation. This embodiment comprehensively considers both the common observation situation and the independent observation situation of the radar pair, enabling a more comprehensive and accurate measurement of the degree of synchronization between the radar pairs in spatial observation.

[0064] S3. Perform steady-state observation relationship analysis on the observation correlation map sequence of the radar detection area to obtain the spatial steady-state feature points of the radar detection area. Update the feature points of the spatial steady-state feature library of the radar detection area based on the spatial steady-state feature points of the radar detection area.

[0065] Specifically, a steady-state observation relationship analysis is performed on the observation correlation map sequence of the radar detection area. The specific process is as follows:

[0066] In the observation correlation map sequence of radar detection areas arranged in time series, the correlation degree of each radar pair in each monitoring period is extracted to form the correlation degree time series of the radar pair. The correlation degree duration threshold, correlation degree variance threshold and continuous period number threshold are set.

[0067] It should be added that the specific methods for setting the correlation persistence threshold, correlation variance threshold, and continuous period number threshold are as follows: Those skilled in the art, through analysis of historical radar observation data, have found that the average correlation of radar pairs generated by real targets is approximately 0.3, while the average correlation of radar pairs generated by fixed spurious sources, such as communication towers, is approximately 0.85. To effectively distinguish between the two and tolerate a certain amount of observation noise, the correlation persistence threshold is usually set between 0.6 and 0.8. In a specific example, the midpoint 0.7 is taken as the correlation persistence threshold. Values ​​below this indicate that the synchronous observations of the two radars are too accidental or weak to form a steady-state observation relationship. The correlation variance threshold is used to ensure the stability of synchronization. Those skilled in the art have found through analysis that fixed spurious sources, due to their fixed reflector positions, have smaller fluctuations in radar observation correlation, with an average variance of approximately 0.02. Moving targets, due to continuous position changes, have an average correlation variance of approximately 0.15. This threshold is used to screen for targets with stable observation synchronization. In a specific example, the correlation variance threshold for radar pairs can be set to 0.05. A variance higher than this indicates drastic fluctuations in synchronization, which may originate from maneuvering targets or transient interference, rather than fixed false sources. The continuous period number threshold is used to require that the steady-state observation relationship be continuous in time. Those skilled in the art, through analysis of system processing delays and false track generation patterns, have found that brief synchronizations of less than 5 periods are mostly accidental overlaps, while synchronizations of more than 10 periods may miss the timely identification of transient false sources. Therefore, the continuous period number threshold is generally set to 5 to 10 monitoring periods. In a specific example, the median value of 8 periods is taken as the continuous period number threshold, ensuring that the discovered steady-state pattern has temporal continuity and excluding brief coincidences. The specific values ​​of these thresholds are determined and solidified after system deployment by analyzing historical real track data and observation data generated by known fixed false sources such as communication towers, with the optimization goal of achieving a preset balance between false alarm rate and missed detection rate.

[0068] A steady-state observation relationship is determined to exist between radar pairs when the correlation time series of a radar pair meets the following criteria: within a continuous monitoring period of not less than the threshold number of consecutive periods, the correlation degree of the radar pair is always higher than the correlation duration threshold, and the variance of the correlation degree of the radar pair is lower than the correlation variance threshold within the monitoring period. This criterion comprehensively considers the persistence and stability of the correlation degree. Only when the radar pair maintains a high and stable correlation degree over multiple continuous monitoring periods is a steady-state observation relationship determined. This helps to accurately identify radar pairs that truly have stable observation characteristics, avoid misjudgments, and thus provide a reliable basis for subsequently determining spatial steady-state feature points.

[0069] The false track identification method based on multi-radar detection also includes a method for initially constructing a spatial steady-state feature database of the radar detection area, which includes:

[0070] Steady-state observation relationship analysis is performed on the observation correlation map sequence of the radar detection area within the historical monitoring period to obtain the historical spatial steady-state feature points of the radar detection area. The feature labels of these historical spatial steady-state feature points are recorded, and a spatial steady-state feature library of the radar detection area is constructed. The spatial steady-state feature library of the radar detection area contains the feature labels of the historical spatial steady-state feature points of the radar detection area. The feature labels of the historical spatial steady-state feature points include the center coordinates of the historical spatial steady-state feature points, the spatial distribution range, the corresponding radar pair identifier, and the initial confidence value.

[0071] It should be added that the confidence value of a spatial steady-state feature point represents the probability estimate that the spatial location corresponding to the feature point actually has a fixed false signal source that can be stably and repeatedly observed by a specific radar during continuous monitoring. For newly identified spatial steady-state feature points added to the spatial steady-state feature library of the radar detection area, the confidence value in the feature label of the spatial steady-state feature point will be set to a high preset value, i.e., the initial confidence value. The initial confidence value is not set arbitrarily, but is determined through the following process: First, historical steady-state feature points with similar judgment conditions to the feature point are retrospectively analyzed, and the proportion and stability of these historical steady-state feature points that were confirmed as real fixed false sources in subsequent long-term monitoring are analyzed. Then, domain experts combine... The analysis results, the stringency of the correlation persistence and stability criteria satisfied by the current feature point, and the corresponding radar performance parameters are comprehensively scored. The average of the scores from multiple experts is taken to obtain the initial confidence value of the feature point. In current engineering practice, according to the above process, for newly added feature points that meet the judgment conditions, the initial confidence value obtained after processing is usually set to 0.8. This means that based on historical experience and expert consensus, it is initially identified with an 80% confidence level that there is a potential fixed false source at this location that needs to be continuously monitored. Starting from this point, the confidence level will be dynamically decayed or enhanced and updated in subsequent continuous monitoring based on whether the feature point is repeatedly confirmed or gradually refuted by new observation data.

[0072] Specifically, the feature points of the spatial steady-state feature database of the radar detection area are updated. The specific process is as follows:

[0073] For radar pairs identified as having a steady-state observation relationship in the observation association sequence of the radar detection area, the following operations are performed: All original point data uploaded by the radar pair within the continuous monitoring period upon which the determination is based are traced back; three-dimensional spatial grid cells simultaneously observed by the radar pair within a monitoring period exceeding a preset proportion are selected; the selected three-dimensional spatial grid cells are clustered in three-dimensional spatial coordinates using the K-Means clustering algorithm; each cluster forming a spatially continuous and dense set of cells is defined as a spatial steady-state feature point, denoted as the spatial steady-state feature point of the radar detection area; and the center coordinates, spatial distribution range, and corresponding radar pair identifier of the spatial steady-state feature point are recorded; thereby determining whether the spatial steady-state feature point of the radar detection area exists in the spatial steady-state feature database.

[0074] To determine whether a spatial steady-state feature point in the radar detection area exists in the spatial steady-state feature library, the specific method is as follows: compare the center coordinates, spatial distribution range, and corresponding radar pair identifier of the spatial steady-state feature point in the radar detection area with each spatial steady-state feature point in the spatial steady-state feature library using feature matching.

[0075] First, spatial location feature matching and comparison are performed. The three-dimensional Euclidean distance between the center coordinates of the spatial steady-state feature point and the center coordinates of each spatial steady-state feature point in the spatial steady-state feature library is calculated. At the same time, the volume overlap ratio between the spatial distribution range of the spatial steady-state feature point and the spatial distribution range of each spatial steady-state feature point in the spatial steady-state feature library is calculated. The volume overlap ratio is the ratio of the volume of the intersection of the two spatial distribution ranges to the union volume of the two. If and only if the three-dimensional Euclidean distance is less than or equal to a set position deviation distance threshold and the volume overlap ratio is greater than or equal to a preset spatial range overlap threshold, the two spatial steady-state feature points are determined to match in spatial location, and then radar identification matching and comparison are performed. Otherwise, it is determined that the spatial steady-state feature point of the radar detection area does not exist in the spatial steady-state feature library.

[0076] It should be noted that the position deviation distance threshold is used to tolerate the inherent measurement errors of the radar system and the slight center shift that may occur due to different periodic clustering analyses. Those skilled in the art determine this threshold based on the statistical positioning accuracy value of the radar within the detection area. In order to allow reasonable measurement fluctuations while ensuring matching reliability, in a specific example, the position deviation distance threshold is set to twice the statistical positioning accuracy value. For example, if the statistical positioning accuracy value is 5 meters, then the threshold is set to 10 meters. The spatial range overlap threshold is used to quantify the consistency of the regions represented by two spatial steady-state feature points. In a specific example, those skilled in the art set the spatial range overlap threshold to 0.5 in order to balance the strictness and fault tolerance of the matching. If the threshold is too low, different regions with only a small overlap in space may be misjudged as the same feature point; if the threshold is too high, feature points belonging to the same physical source may not be matched due to slight differences in the clustering boundary. That is, the volume overlap of the two spatial distribution ranges must reach at least 50% of their total volume before spatial position matching can be determined. This value is intended to ensure that the spatial regions represented by the feature points have substantial overlap.

[0077] The radar pair identifier matching and comparison process is as follows: compare the radar pair identifier corresponding to the spatial steady-state feature point with the radar pair identifier recorded in the feature label of each spatial steady-state feature point in the spatial steady-state feature library. If and only if the two radar pair source identifiers are completely consistent, it is determined that the spatial steady-state feature point of the radar detection area exists in the spatial steady-state feature library.

[0078] If the spatial steady-state feature point does not exist in the spatial steady-state feature library, the feature label of the spatial steady-state feature point is recorded. The feature label of the spatial steady-state feature point includes the center coordinates of the spatial steady-state feature point, the spatial distribution range, the corresponding radar pair identifier and the initial confidence value, and the feature label of the spatial steady-state feature point is added to the spatial steady-state feature library of the radar detection area.

[0079] The center coordinates of the spatial steady-state feature points are obtained by clustering the coordinates of the selected three-dimensional spatial grid units using the K-Means clustering algorithm. The mean center coordinates of all units in the cluster are calculated directly by the algorithm. The spatial distribution range of the spatial steady-state feature points is determined by calculating the minimum and maximum coordinates of these units in the three-dimensional coordinate system of the unified geographic coordinate system. The minimum bounding three-dimensional cube that can completely enclose all units in the cluster is determined by calculating the minimum and maximum coordinates of these units in the east-west, north-south, and height dimensions.

[0080] The three-dimensional spatial grid cell simultaneously observed by the radar refers to the observation vector element corresponding to the spatial grid cell within the same monitoring period, where both radars assign observation markers. By tracing back the original point data and filtering out the simultaneously observed grid cells, and then performing cluster analysis to determine the spatial steady-state feature points, it is possible to accurately locate spatial positions with stable observation characteristics in the radar detection area. These spatial steady-state feature points are of great significance for understanding the characteristics of the radar detection area and identifying potential fixed false signal sources. At the same time, an initial confidence level is assigned to the newly added feature points, laying the foundation for subsequent reliability assessment and dynamic updates of the feature points.

[0081] It should be added that the preset ratio is used to screen spatial areas that are stably and jointly observed by the radar pair. Those skilled in the art, through analysis of historical observation data, have found that the average proportion of accidental spatial overlap caused by moving targets or transient interference being simultaneously observed by the same radar pair is less than 50%, while the average proportion of fixed false sources is greater than 90%. To balance detection sensitivity and false alarm rate, in a specific example, the preset ratio is set to 0.75, meaning that a three-dimensional spatial grid cell must be simultaneously observed by the radar pair for at least 75% of the monitoring period to be considered a potential fixed false source candidate. This preset ratio aims to ensure that the located feature points have high observation repeatability, thereby effectively distinguishing them from accidental spatial overlap caused by moving targets or transient interference. This specific value is determined based on statistical analysis of historical observation data, with the goal of balancing detection sensitivity and false alarm rate.

[0082] For the feature labels of existing spatial steady-state feature points in the spatial steady-state feature library, if the radar pair associated with the spatial steady-state feature point continues to be determined to have a steady-state observation relationship within the current monitoring period, a confidence increase operation is performed. The confidence value in the feature label of the spatial steady-state feature point is increased by a preset confidence increase step size to obtain the confidence value after the increase operation. A confidence upper limit threshold is set. If the confidence value after the increase operation is higher than the confidence upper limit threshold, the confidence value in the feature label of the spatial steady-state feature point is fixed to the confidence upper limit threshold. Otherwise, the confidence value in the feature label of the spatial steady-state feature point is updated to the confidence value after the increase operation. This dynamic increase method can strengthen the identification weight of stable false source points that have been repeatedly verified over time, making the system's accumulated cognition more stable and reliable. At the same time, by setting a confidence upper limit, the confidence value is prevented from growing indefinitely.

[0083] If the radar pair associated with the spatial steady-state feature point is no longer considered to have a steady-state observation relationship within the current monitoring period, a confidence reduction operation is performed. This reduces the confidence value in the feature label of the spatial steady-state feature point by a preset confidence reduction step size, resulting in a confidence value after the reduction operation. A confidence removal threshold is then set. If the confidence value after the reduction operation is lower than the confidence removal threshold, the feature label of the spatial steady-state feature point is deleted from the spatial steady-state feature library. Otherwise, the confidence value in the feature label of the spatial steady-state feature point is updated to the confidence value after the reduction operation. This reduction and removal method enables the system to autonomously eliminate outdated feature points that are formed accidentally, exist for a short period of time, or have changed, ensuring the timeliness and cleanliness of the spatial steady-state feature library. It effectively prevents historical invalid information from interfering with the current analysis and realizes the self-purification and adaptive updating of the spatial steady-state feature library.

[0084] It should be added that the preset confidence increase step size, confidence upper limit threshold, preset confidence decrease step size, and confidence removal threshold are set as follows: In a specific example, in order to enable the feature library to both robustly accumulate reliable knowledge and sensitively remove invalid or outdated information, those skilled in the art have determined the following parameter combination based on historical data statistics: the confidence increase step size is set to 0.05, that is, whenever the radar pair corresponding to the feature point is determined to have a steady-state observation relationship, its confidence increases by 0.05; the confidence upper limit threshold is set to 0.95, when... Once the confidence level reaches this cumulative value, it will not increase further to ensure the stability of the system's cognition. The confidence level reduction step size can be set to 0.10, meaning that when the radar pair no longer satisfies the steady-state observation relationship, the confidence level decreases at a rate greater than the increase. The confidence level removal threshold is set to 0.30. When the confidence level drops below this value due to continuous failure, the feature point record will be deleted from the database. This combination of parameters—small, slow increases, large, rapid decreases, and low-level clearing—is specifically determined based on historical data statistics. It aims to enable the feature database to both robustly accumulate reliable knowledge and sensitively remove invalid or outdated information.

[0085] In this implementation plan, by analyzing the sequence of observation correlation diagrams, steady-state characteristics in the observation synchronization relationship of the radar network can be extracted, and a spatial steady-state characteristic database can be constructed and updated, providing an important reference for the subsequent judgment of track risk level.

[0086] S4. Collect radar detection area track data generated by the data fusion center, combine it with the spatial steady-state feature library of the radar detection area, perform logical anomaly detection and feature matching analysis on the radar detection area track, and obtain the risk level of the radar detection area track.

[0087] Specifically, the process of collecting radar detection area trajectory data generated by the data fusion center is as follows:

[0088] The data fusion center receives the original point data from the radar and performs time-series correlation matching to obtain radar detection area track data. The radar detection area track data includes radar detection area track, track identifier, and a sequence of point traces that constitute the radar detection area track arranged in chronological order. The point trace sequence includes the timestamp, three-dimensional spatial coordinates, and source identifier of the point trace.

[0089] The radar detection area track data is obtained by performing time-series correlation matching. The specific process is as follows:

[0090] The data fusion center receives raw radar point data, arranges the raw points in the radar raw point data in chronological order according to timestamps, and then performs correlation matching on raw points with the same source identifier and adjacent three-dimensional spatial coordinates. Based on the correlation matching raw points, a point sequence is constructed, and then a radar detection area track is generated according to the point sequence. At the same time, each radar detection area track is assigned a unique track identifier. Finally, radar detection area track data containing radar detection area tracks, track identifiers, and the point sequence that constitutes the track arranged in chronological order is obtained. The asynchronous and discrete raw point data of each radar are transformed and organized into track data with clear spatiotemporal correlation, providing a reliable data foundation for subsequent track-level risk analysis.

[0091] Specifically, logical anomaly detection and feature matching analysis are performed on the flight tracks in the radar detection area. The specific process is as follows:

[0092] Step a1: Analyze the radar detection area track data to obtain the key features of the radar detection area track. The key features of the radar detection area track include the radar composition features and the spatial residing core area of ​​the radar detection area track. The set of source identifiers of all points constituting the radar detection area track is used as the radar composition features of the radar detection area track. Using a spatial density clustering algorithm, such as DBSCAN, the clustering area of ​​all points constituting the radar detection area track in three-dimensional space is analyzed as the spatial residing core area of ​​the radar detection area track. The radar composition features and spatial residing core areas of all radar detection area tracks are obtained by traversing the entire process.

[0093] Step a2: Based on the radar detection area track data, perform logic anomaly detection on the radar detection area track, including timing anomaly analysis and motion anomaly analysis.

[0094] Temporal anomaly analysis and motion anomaly analysis are performed on the tracks in the radar detection area to obtain the temporal source anomaly determination results and motion anomaly determination results of the tracks in the radar detection area, and then the risk level of the tracks in the radar detection area is obtained.

[0095] Perform timing anomaly analysis on the flight tracks in the radar detection area. The specific analysis process includes:

[0096] Based on the timestamps and source identifiers in the track's point sequence, the theoretical minimum time required for the monitored target to switch between adjacent points from different radar sources is calculated and compared with the actual time interval. If the actual time interval is less than the theoretical minimum time, the radar switch is determined to be a physically unreachable switch. The number of physically unreachable switches in the track is counted. If the number of physically unreachable switches in the track is greater than or equal to a preset anomaly threshold, the track is determined to have a time-series source anomaly; if the number of physically unreachable switches is less than the preset anomaly threshold, the track is determined to have a time-series source anomaly. If a constant threshold is reached, the trajectory is determined to be free of temporal anomalies. This yields the temporal anomaly determination result for the trajectory in the radar detection area. The temporal anomaly determination result for the trajectory in the radar detection area includes the presence of temporal anomalies and the absence of temporal anomalies. Temporal anomaly analysis examines whether the trajectory point sequence violates the physical constraints of the target's motion capability in terms of temporal logic. It can effectively analyze logical contradictions formed by the incorrect association of different real target trajectory segments on the timeline, providing a key criterion for identifying false trajectories with coincidental trajectories without a fixed physical source.

[0097] It should be added that the preset anomaly threshold is used to identify obvious abnormal jump patterns in the track point sequence. Those skilled in the art, through statistical analysis of historical track data, have found that in real tracks, accidental physical unreachable jumps caused by observation noise or synchronization errors occur on average about 1.2 times per 100 radar jumps, while systematic logical contradictions formed by the incorrect association of multiple unrelated trajectories occur on average about 4.5 times per 10 jumps. To effectively capture systematic contradictions while tolerating accidental errors, in a specific example, the anomaly threshold can be set to 3 times. That is, once at least 3 physically unreachable radar jumps occur, the track is determined to have a temporal traceability anomaly. This anomaly threshold setting is based on engineering experience, aiming to tolerate accidental jumps caused by a small amount of observation noise or synchronization errors, while sensitively capturing systematic logical contradictions caused by the incorrect association of multiple unrelated trajectories. The specific value of the anomaly threshold is determined based on statistical analysis of historical track data under typical mission scenarios, and after weighing the acceptable false positive rate against the false negative rate.

[0098] The theoretical minimum time required for a monitored target to switch between adjacent points from different radar sources is calculated. The calculation method is based on the difference in three-dimensional spatial coordinates between adjacent points of the monitored target, combined with the maximum speed of the monitored target within the radar detection area, to obtain the theoretical minimum time.

[0099] It should be added that the maximum speed of the monitored target within the radar detection area is used to set a reasonable physical speed boundary to determine whether the target can jump between radars. In a specific embodiment, those skilled in the art, during the initial deployment or learning phase, collect and process a large number of tracks generated by verified real targets, calculate their speed distribution, select the highest speed value as the initial estimate of the maximum speed, and fine-tune and solidify it according to the specific mission scenario, such as the performance upper limit of common targets in urban low-altitude UAV management, thereby providing a reasonable and reliable speed benchmark for physical reachability judgment.

[0100] Perform motion anomaly analysis on the flight tracks in the radar detection area. The specific analysis process includes:

[0101] Based on the radar composition characteristics of the radar detection area tracks, the point sequence of the radar detection area tracks is back-projected onto the corresponding coordinate system of each radar. The corresponding motion trajectory of each radar is then fitted, and the deviations in velocity, acceleration, and heading changes between different trajectories are calculated. This yields the motion anomaly determination result for the radar detection area tracks. If the velocity deviation value exceeds a preset velocity anomaly deviation threshold, or the acceleration deviation value exceeds a preset acceleration anomaly deviation threshold, or the heading deviation value exceeds a preset heading anomaly deviation threshold, then the motion anomaly determination result for that radar detection area track is marked as having a motion anomaly; otherwise, the radar detection area track is marked as not having a motion anomaly. The motion anomaly determination result of the radar detection area track is marked as having no motion anomaly, thus obtaining the motion anomaly determination result of the radar detection area track. The motion anomaly determination result of the radar detection area track includes the presence of motion anomaly and the absence of motion anomaly. The motion anomaly analysis starts from the perspective of kinematic consistency. By examining whether the motion state of the same target under different radar local observation perspectives is logical, it can keenly discover tracks that seem reasonable under the global fusion perspective but are contradictory under the local observation perspective due to information fusion errors or complex interference. This enhances the ability to detect more concealed systematic misjudgment tracks.

[0102] It should be added that the specific methods for setting the preset speed anomaly deviation threshold, acceleration anomaly deviation threshold, and heading anomaly deviation threshold are as follows: those skilled in the art determine these thresholds based on the normal motion parameter range of the targets monitored by the system, such as civil aircraft and general aviation aircraft, combined with the known speed and direction measurement accuracy of each radar, by statistically analyzing the maximum reasonable deviation in historical normal flight track data. In a specific example, the speed anomaly deviation threshold can be set to 10 m / s, that is, when the speed difference fitted from different radar data exceeds this value, it is considered kinematic inconsistency; the acceleration anomaly deviation threshold can be set to 5 m / s², used to identify abnormal acceleration that violates normal maneuverability; the heading anomaly deviation threshold can be set to 15 degrees, used to detect contradictions in the target's heading. The specific values ​​of these thresholds are determined based on the normal motion parameter range of the targets monitored by the system, such as civil aircraft and general aviation aircraft, combined with the known speed and direction measurement accuracy of each radar, by statistically analyzing the maximum reasonable deviation in historical normal flight track data.

[0103] If the timing source anomaly determination result of the radar detection area track indicates the presence of a timing source anomaly, or the motion anomaly determination result of the radar detection area track indicates the presence of a motion anomaly, then the radar detection area track is assigned a high-risk level; otherwise, the radar detection area track is assigned a normal risk level.

[0104] Step a3: Perform feature matching analysis on the key features of the radar detection area track and the feature labels of each spatial steady-state feature point in the spatial steady-state feature library to obtain the source radar matching degree and spatial overlap degree between the radar detection area track and each spatial steady-state feature point.

[0105] The source radar composition characteristics and historical spatial residence core area of ​​the radar detection area track are matched with the source identifiers and spatial distribution ranges of radar pairs associated with each spatial steady-state feature point recorded in the spatial steady-state feature database. The matching degree includes source radar matching degree and spatial overlap degree. The source radar matching degree is the proportion of the total number of identical radar identifiers of the source radar composition characteristics of the radar detection area track and the source identifiers of the radar pairs associated with the spatial steady-state feature points to the total number of radar identifiers in the source radar composition characteristics. The spatial overlap degree is the proportion of the overlap volume between the spatial residence core area of ​​the radar detection area track and the spatial distribution range of the spatial steady-state feature points to the volume of the spatial residence core area. This matching process realizes the accurate association between the upper-layer dynamic track and the lower-layer static false source. By quantitatively calculating the similarity of radar composition and the overlap of spatial position, an objective and measurable standard is provided for judging whether a track may originate from a known fixed false signal source.

[0106] Set a source radar matching degree threshold and a spatial overlap degree threshold. If there is a spatial steady-state feature point, such that the source radar matching degree between the track in the radar detection area and the steady-state feature point is higher than the source radar matching degree threshold and the spatial overlap degree is higher than the spatial overlap degree threshold, then the track in the radar detection area is assigned a high-risk level; otherwise, the track in the radar detection area is assigned a normal risk level.

[0107] It should be added that the specific methods for setting the source radar matching degree threshold and the spatial overlap degree threshold are as follows: those skilled in the art determine these thresholds by performing statistical analysis on verified false and real flight tracks in historical data, with the goal of optimizing the overall recognition performance of the system. In a specific example, the source radar matching degree threshold can be set to 0.65 and the spatial overlap degree threshold can be set to 0.60. That is, when the radar composition matching degree of a flight track with a certain spatial steady-state feature point exceeds 65%, and the distribution range of its spatial core area overlaps with that feature point by more than 60%, it is determined that the flight track is highly correlated with that feature point. This set of thresholds can provide a clear and quantifiable judgment standard for the association between the flight track and the fixed false source. The specific value of the threshold is determined based on the statistical analysis on verified false and real flight tracks in historical data, with the goal of optimizing the overall recognition performance of the system.

[0108] In this implementation plan, a high-risk target screening method based on feature matching and logical analysis is constructed by performing bidirectional cross-validation between the analyzed false source features and the upper-layer fused track, and introducing a spatiotemporal logic consistency test. This method can not only efficiently screen out false targets generated by fixed environmental reflectors, but also effectively identify dynamic logically abnormal targets generated by system misassociations or complex interference, thereby improving the identification range of false tracks.

[0109] S5. Based on the risk level of the radar detection area track, determine the type of radar detection area track, and update the confidence level of the spatial steady-state feature database of the radar detection area.

[0110] Specifically, the process for determining the type of flight track in the radar detection area is as follows:

[0111] Step b1: For radar detection area tracks that are assigned a high-risk level due to the time-series source anomaly determination result or the motion anomaly determination result, the radar detection area track type is determined to be a dynamic anomaly type false track.

[0112] Step b2: For radar detection area tracks that are assigned a high-risk level due to feature matching analysis with the spatial steady-state feature library, the type of radar detection area track is determined to be a fixed false source type false track.

[0113] Step b3: If the radar detection area track simultaneously meets the determination conditions of steps b1 and b2 above, then the radar detection area track is determined to be a composite false track. Such tracks may originate from a complex interference source that can produce stable reflections and whose signal characteristics cause kinematic or temporal contradictions in the fusion system.

[0114] Step b4: Otherwise, the radar detection area is determined to be a real track. This effectively distinguishes real tracks from false tracks, ensuring the reliability of the track judgment results and providing a data basis for subsequent operations based on the track judgment results.

[0115] Specifically, the confidence level of the spatial steady-state feature database of the radar detection area is updated by feedback. The specific process is as follows:

[0116] The data fusion center feeds back the judgment results to the spatial steady-state feature library of the radar detection area. If the radar detection area track is judged to be a fixed false source type false track or a composite false track, the confidence level is increased for the matching spatial steady-state feature points of the radar detection area track in the spatial steady-state feature library.

[0117] If a radar detection area track is determined to be a genuine track, a confidence reduction operation is performed on the matching spatial steady-state feature points of the radar detection area track in the spatial steady-state feature database. This completes the feedback update of the spatial steady-state feature database of the radar detection area. This feedback update enables the spatial steady-state feature database to continuously learn and adapt to the actual detection situation, improving the accuracy and practicality of the spatial steady-state feature database, and further enhancing the entire radar detection system's ability to identify and process false tracks.

[0118] Please see Figure 2 As shown, a second aspect of the present invention provides a false track identification system based on multi-radar detection, comprising:

[0119] The spot data acquisition module is used to deploy data acquisition units on radars in the radar detection area. Within a preset monitoring period, the data acquisition units collect the original spot data of the radar and transmit the original spot data of the radar to the data fusion center.

[0120] The correlation graph generation module is used to receive the original radar point data through the data fusion center, construct the joint observation matrix of the radar detection area, and generate the observation correlation graph sequence of the radar detection area arranged in time series.

[0121] The feature library update module is used to perform steady-state observation relationship analysis on the observation correlation map sequence of the radar detection area, obtain the spatial steady-state feature points of the radar detection area, and update the spatial steady-state feature library of the radar detection area based on the spatial steady-state feature points of the radar detection area.

[0122] The risk analysis module is used to collect radar detection area track data generated by the data fusion center, combine it with the spatial steady-state feature library of the radar detection area, perform logical anomaly detection and feature matching analysis on the radar detection area track, and obtain the risk level of the radar detection area track.

[0123] The judgment and analysis module is used to determine the type of radar detection area track based on the risk level of the radar detection area track, and to update the confidence value of the spatial steady-state feature library of the radar detection area.

[0124] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0125] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A method for identifying false flight paths based on multi-radar detection, characterized in that, include: S1. Deploy a data acquisition unit on the radar in the radar detection area. Within the preset monitoring period, the data acquisition unit collects the original radar point data and transmits the original radar point data to the data fusion center. S2. The data fusion center receives the original radar spot data, constructs a joint observation matrix of the radar detection area, and generates an observation correlation map sequence of the radar detection area arranged in time series. S3. Perform steady-state observation relationship analysis on the observation correlation map sequence of the radar detection area to obtain the spatial steady-state feature points of the radar detection area. Update the feature points of the spatial steady-state feature library of the radar detection area based on the spatial steady-state feature points of the radar detection area. S4. Collect radar detection area trajectory data generated by the data fusion center, combine it with the spatial steady-state feature library of the radar detection area, perform logical anomaly detection and feature matching analysis on the radar detection area trajectory, and obtain the risk level of the radar detection area trajectory. S5. Based on the risk level of the radar detection area track, determine the type of radar detection area track, and update the confidence level of the spatial steady-state feature database of the radar detection area.

2. The false track identification method based on multi-radar detection according to claim 1, characterized in that: The method for constructing the joint observation matrix of the radar detection area includes: After receiving the original point data from the radar, the data fusion center divides the radar detection area into a three-dimensional spatial grid of a preset size. In each monitoring cycle, it generates a corresponding observation vector for each radar. The length of the observation vector is equal to the total number of three-dimensional spatial grid cells, and each vector element of the observation vector corresponds to a three-dimensional spatial grid cell. If at least one original point of the corresponding radar falls into the currently processed three-dimensional spatial grid cell, the vector element corresponding to the three-dimensional spatial grid cell is marked as having an observation label; otherwise, it is marked as having no observation label. The observation vectors of all radars in the monitoring period are combined to form a joint observation matrix of the radar detection area in the corresponding monitoring period. Each row of the joint observation matrix corresponds to a radar according to a predefined radar sorting rule, and each column corresponds to a three-dimensional spatial grid cell according to a predefined spatial grid coding rule.

3. The false track identification method based on multi-radar detection according to claim 2, characterized in that: The method for generating a time-series sequence of observation correlation maps of radar detection areas includes: Based on the joint observation matrix of the radar detection area, the correlation degree between the observation vectors of all different radar pairs in the joint observation matrix is ​​calculated to construct an observation correlation map. The correlation degree is used to quantify the synchronization degree of the two radars in spatial observation. The radar pair is a combination of any two different radars in the joint observation matrix of the radar detection area. The observation correlation map is constructed for each monitoring cycle to generate a sequence of observation correlation maps of the radar detection area arranged in time series.

4. The false track identification method based on multi-radar detection according to claim 3, characterized in that: The method for analyzing the steady-state observation relationship of the observation correlation map sequence of the radar detection area includes: In the observation correlation map sequence of radar detection area arranged in time series, the correlation degree of each radar pair in each monitoring period is extracted to form the correlation degree time series of the corresponding radar pair. The correlation degree duration threshold, correlation degree variance threshold and continuous period number threshold are set. When a radar pair has a correlation time series that satisfies the following criteria, it is determined that there is a steady-state observation relationship between the radar pairs. The criteria are: within a continuous monitoring period of not less than the number of continuous periods threshold, the correlation degree of the radar pairs is higher than the correlation degree duration threshold, and the variance of the correlation degree of the radar pairs is lower than the correlation degree variance threshold within the continuous monitoring period.

5. The false track identification method based on multi-radar detection according to claim 4, characterized in that: The method for updating feature points in the spatial steady-state feature library of the radar detection area includes: Analyze the radar pairs identified as having a steady-state observation relationship in the observation association sequence of the radar detection area to obtain spatial steady-state feature points of the radar detection area. Record the center coordinates, spatial distribution range, and corresponding radar pair identifier of the spatial steady-state feature points. Determine whether the spatial steady-state feature points exist in the spatial steady-state feature library. If the spatial steady-state feature points do not exist in the spatial steady-state feature library, record the feature labels of the spatial steady-state feature points. The feature labels of the spatial steady-state feature points include the center coordinates, spatial distribution range, corresponding radar pair identifier, and initial confidence value of the spatial steady-state feature points. Add the feature labels of the spatial steady-state feature points to the spatial steady-state feature library of the radar detection area. For the feature labels of existing spatial steady-state feature points in the spatial steady-state feature library, if the radar pair associated with the spatial steady-state feature point is still determined to have a steady-state observation relationship in the current monitoring period, then the confidence increase operation is performed, and the confidence value in the feature label of the corresponding spatial steady-state feature point is updated. If the radar pair associated with the spatial steady-state feature point is no longer determined to have a steady-state observation relationship within the current monitoring period, a confidence reduction operation is performed, and the confidence value in the feature label of the corresponding spatial steady-state feature point is updated.

6. The false track identification method based on multi-radar detection according to claim 1, characterized in that: The method for collecting radar detection area track data generated by the data fusion center includes: The data fusion center receives the original point data from the radar and performs time-series correlation matching to obtain radar detection area track data. The radar detection area track data includes radar detection area track, track identifier, and point sequence that constitutes the radar detection area track arranged in chronological order.

7. The false track identification method based on multi-radar detection according to claim 6, characterized in that: The method for performing logical anomaly detection and feature matching analysis on flight tracks in the radar detection area includes: Step a1: Analyze the radar detection area track data to obtain the key features of the radar detection area track. The key features of the radar detection area track include the radar composition features and the space stationary core area of ​​the radar detection area track. Step a2: Based on the radar detection area track data, perform logical anomaly detection on the radar detection area track to obtain the time-series source anomaly judgment result and motion anomaly judgment result of the radar detection area track, and process to obtain the risk level of the radar detection area track; Step a3: Perform feature matching analysis between the key features of the radar detection area track and the feature labels of each spatial steady-state feature point in the spatial steady-state feature library to obtain the source radar matching degree and spatial overlap degree between the radar detection area track and each spatial steady-state feature point, and process to obtain the risk level of the radar detection area track.

8. The false track identification method based on multi-radar detection according to claim 7, characterized in that: The method for determining the type of flight track in the radar detection area includes: Step b1: For radar detection area tracks that are assigned a high-risk level due to the time-series source anomaly determination result or the motion anomaly determination result, the radar detection area track type is determined to be a dynamic anomaly type false track. Step b2: For radar detection area tracks that are assigned a high-risk level due to feature matching analysis with the spatial steady-state feature library, the type of radar detection area track is determined to be a fixed false source type false track; Step b3: If the radar detection area track simultaneously meets the determination conditions of step b1 and step b2, then the radar detection area track is determined to be a composite false track. Step b4: Otherwise, determine that the type of the track in the radar detection area is a real track.

9. The method for identifying false flight paths based on multi-radar detection according to claim 1, characterized in that, It also includes an initial method for constructing a spatial steady-state feature library of the radar detection area, including: Steady-state observation relationship analysis is performed on the observation correlation map sequence of the radar detection area within the historical monitoring period to obtain the historical spatial steady-state feature points of the radar detection area. The feature labels of the historical spatial steady-state feature points are recorded to construct a spatial steady-state feature library of the radar detection area, which contains the feature labels of the historical spatial steady-state feature points of the radar detection area.

10. A false track identification system based on multi-radar detection, used to implement the false track identification method based on multi-radar detection as described in any one of claims 1-9, characterized in that, include: The spot data acquisition module is used to deploy data acquisition units on radars in the radar detection area. Within a preset monitoring period, the data acquisition units collect the original spot data of the radar and transmit the original spot data of the radar to the data fusion center. The correlation graph generation module is used to receive the original radar point data through the data fusion center, construct the joint observation matrix of the radar detection area, and generate the observation correlation graph sequence of the radar detection area arranged in time series. The feature library update module is used to perform steady-state observation relationship analysis on the observation correlation map sequence of the radar detection area, obtain the spatial steady-state feature points of the radar detection area, and update the spatial steady-state feature library of the radar detection area based on the spatial steady-state feature points of the radar detection area. The risk analysis module is used to collect radar detection area track data generated by the data fusion center, and combine it with the spatial steady-state feature library of the radar detection area to perform logical anomaly detection and feature matching analysis on the radar detection area track to obtain the risk level of the radar detection area track. The judgment and analysis module is used to determine the type of radar detection area track based on the risk level of the radar detection area track, and to update the confidence value of the spatial steady-state feature library of the radar detection area.

Images