Efficient track association method based on double threshold rejection strategy

Abstract

The present invention discloses an efficient track association method based on a double threshold rejection strategy, including receiving and preprocessing track information reported by each radar, unifying the parameters of speed and heading angle at the same time in the same coordinate system; obtaining the speed information of the two radars to form a speed vector, calculating the speed screening parameter matrix and comparing it with the speed error threshold, completing the first round of rejection calculation; taking out the heading angles of the corresponding tracks to form the heading vectors of the two radars, calculating the heading screening parameter matrix according to the heading vector and comparing it with the heading error threshold, completing the heading rejection calculation; and then using a double threshold track association algorithm to complete track association. The present invention can be processed in parallel through speed and heading rejection operations. After two-step rejection and screening, the remaining track combinations have been reduced by 70% to 80%. Then, using a track association algorithm, completing track association can quickly eliminate speed and heading mismatched track pairs, greatly reducing the amount of calculation and improving track association efficiency.

Description

Technical Field

[0001] The invention relates to the technical field of radar information detection, and in particular to an efficient track association method based on a double threshold rejection strategy. Background Art

[0002] Radar is the core of the battlefield and an essential means of modern warfare. Without radar, other means of warfare will not be implemented. For this reason, countermeasures for radar detection are emerging in an endless stream, and technologies such as electronic deception, electronic interference, and low, small, and slow are becoming more and more mature. Single radar detection can no longer meet battlefield needs, and the selection of radar networking that organically combines different frequency bands, working modes, polarization modes, and systems to form a radar network detection group has become the direction of technical research and practice. The distributed radars in the region are connected and communicated through the network, and the command center will fuse the collected detection information to eliminate false alarms, associate the same target track, and fuse a track with higher credibility and accuracy to maximize the radar efficiency.

[0003] In order to solve the problem of multiple radars reporting the same target track information in radar network detection, a track association algorithm is needed to complete the association of the same target track from different reporting sources. Commonly used track association methods include weighted track association algorithm, statistical double threshold track association algorithm, nearest neighbor track association algorithm, etc. Among them, the weighted track association algorithm is a statistical weighting of the state estimation error. When there are many targets or bifurcated tracks, the weighted method will have wrong or missed associations. The statistical double threshold track association algorithm uses the secondary threshold method to greatly reduce the missed association phenomenon and the wrong association phenomenon of crossed and bifurcated tracks, improve the association quality, but at the same time increase the design requirements of the model. The aforementioned association algorithms all first perform association calculations on each point track of each radar track, then complete the association based on a certain rule, and then use some strategies for screening. However, since each radar reports a large number of tracks, the comparison of each track point will take up a considerable amount of calculation and calculation time, thereby affecting the efficiency of track fusion. Summary of the invention

[0004] The purpose of the present invention is to provide an efficient track association method based on a double threshold rejection strategy, which uses speed and heading angle as the basis for information screening, first filters a large amount of information through lightweight calculations, and then completes the track association through an association algorithm, so as to solve the problems of large amount of calculation and low efficiency brought about by directly using a track association algorithm for track association proposed in the above background technology. This method will greatly reduce the amount of calculation of point track association in track association, improve track association efficiency, and improve track fusion performance.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The efficient track association method based on the double threshold rejection strategy includes the following steps:

[0007] Step 1: Establish a radar network detection system, and the fusion center receives the track information reported by each radar; complete radar track preprocessing, which at least includes outlier removal and time alignment, and align the spatial coordinates to the geocentric rectangular coordinate system, and unify the parameters of speed and heading angle at the same time in the same coordinate system;

[0008] Step 2: Obtain the speed information of the two radars R1 and R2, form speed vectors V1 and V2, and calculate the speed screening parameter matrix V according to the speed vectors p , and then the speed screening parameter matrix V p The matrix elements in One by one with the speed error threshold VT u VT d Compare, if it is within the interval, it is considered that the speed meets the element, and it can enter the next round of screening, and so on to complete the first round of elimination calculation, and obtain the speed elimination matrix V r ;

[0009] Step 3: Eliminate the matrix V from the velocity r Find the element whose parameter is 1 Pick The corner mark ij, where i corresponds to the track number of radar R1 and j corresponds to the track number of radar R2, takes out the heading angle of the corresponding track to form the heading vectors A1 and A2 of radars R1 and R2, and calculates the heading screening parameter matrix A according to the heading vectors p , and then compared with the heading error threshold AT u , AT d Compare, if it is within the interval, it is considered that the heading meets the element, and so on, and finally complete the heading elimination calculation to obtain the heading elimination matrix A r ;

[0010] Step 4: After two threshold eliminations based on steps 2 and 3, the heading elimination matrix A r Elements with parameter 1 The subscript ij is called the track candidate mapping group, where i corresponds to the track number of radar R1 and j corresponds to the track number of radar R2. The double threshold track association algorithm is used to complete the track association.

[0011] A further technical solution is: in step 1, the radar track is preprocessed, and the inter-radar data is time-aligned using interpolation and extrapolation methods.

[0012] A further technical solution is: in step 2, the velocity elimination matrix V is obtained by calculating the velocity vectors of the two radars R1 and R2. r, the specific steps are as follows:

[0013] a1) First, obtain the speed information of all reported tracks of radars R1 and R2 to form speed vectors V1 and V2;

[0014] a2) Calculate the transposed matrix of the velocity vector V1

[0015] a3) Multiply the velocity vector V2 by the m-order unit matrix E and convert it into a square matrix V 2E ;

[0016] a4) Calculate V 2E The inverse matrix

[0017] a5) Converted into a row vector, V 2v ;

[0018] a6) Calculate the speed screening parameter matrix V p ,in,

[0019] a7) The speed screening parameter matrix V p The matrix elements in One by one with the speed error threshold VT u and VT d Compare and remove the speed matrix V r Assign values ​​to each element;

[0020] like:

[0021] or but

[0022] like:

[0023] and but

[0024] Repeat the speed error threshold comparison operation until all matrix elements After the comparison is completed, the velocity elimination matrix V is finally obtained. r .

[0025] A further technical solution is: in step 3, the heading elimination matrix A is calculated from the heading vectors of radars R1 and R2 r , the specific steps are as follows:

[0026] b1) First, obtain the heading angles of radars R1 and R2 after velocity elimination to form heading vectors A1 and A2;

[0027] b2) Calculate the transposed matrix of the heading vector A1

[0028] b3) Multiply the heading vector A2 by the q-order unit matrix E and convert it into a square matrix A 2E ;

[0029] b4) Calculate A 2E The inverse matrix

[0030] b5) Converted into a row vector, A 2v ;

[0031] b6) Calculate the heading screening parameter matrix A p ,in,

[0032] b7), put A p The matrix elements in One by one with the heading error threshold AT u and AT d Compare and eliminate the heading matrix A r Assign values ​​to each element;

[0033] like:

[0034] or but

[0035] like:

[0036] and but

[0037] Repeat the heading error threshold comparison operation until all matrix elements After the comparison is completed, the heading elimination matrix A is finally obtained. r .

[0038] A further technical solution is: in step 4, a double-threshold track association algorithm is used to complete track association, and the specific steps are as follows:

[0039] c1) Take out the track of the candidate track mapping group ij, assuming that the filtering value of the radar R1 for the i-th track at time t is And the error covariance of radar R1 for the ith track at time t is At this point, since the time-space calibration has been completed in step 1, it can be calculated directly;

[0040] c2) Assume that the state estimation errors of the two radar stations for the same target are statistically independent, and there exist H0 and H1 that satisfy the following conditions:

[0041] H0: and is the track estimate of the same target;

[0042] H2: and It is not the track estimate of the same target;

[0043] c3) At time t, for the R estimation error samples from two local nodes, firstly calculate them one by one based on χ 2 The distribution threshold is used for hypothesis testing. If H0 is accepted, the counter is incremented by 1. Otherwise, the counter value remains unchanged. The test statistic using the weighted method is calculated point by point:

[0044]

[0045] Among them, α ij (t) is the correlation coefficient between the track i of radar R1 and the track j of radar R2; t is time; is the filtering value of radar R2 for the jth track at time t; is the error covariance of radar R2 for the jth track at time t; T is the matrix transpose calculation;

[0046] When α ij When (t)≤σ, the counter M is incremented by 1; if: M≥L, it is determined that the radar R1 track i and the radar R2 track j are the observation values ​​of the same track. σ and L are two-step threshold parameters, which are taken according to the actual radar conditions in the test phase. Repeat the above steps to complete the correlation calculation of all tracks between radar R1 and radar R2.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] First, the present invention vectorizes the velocity components of the target tracks reported by the radar that need to be associated, and removes the track association pairs with mismatched velocities through matrix operations. By screening the velocity components, the track pairs that may be on the same track are set to the corresponding matrix element position 1, and the matrix elements set to 1 identify the track pairs that may be associated, which will greatly reduce the amount of calculation for subsequent associations.

[0049] Second, while filtering the speed matrix, the heading elimination operation is performed in parallel. The heading angles of all tracks of the two radars are taken out and vectorized. The heading components are filtered again through matrix operation to eliminate tracks with mismatched heading dimensions, and the range of associated tracks is further narrowed, further reducing the amount of calculation for the associated operation and improving the efficiency of the association.

[0050] Third, after eliminating the two-step calculation based on speed and heading, the track association is completed through the track association algorithm. This not only reduces the previous track association process of performing one-by-one association operations on all point tracks of all tracks, but also uses the two major association parameters of heading and speed as association factors, which greatly improves the association accuracy and improves the track fusion efficiency and accuracy of multi-radar network detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flow chart of an efficient track association method based on a double threshold rejection strategy according to an embodiment of the present invention;

[0052] Figure 2 is the speed elimination matrix V in the embodiment of the present invention r The acquisition process;

[0053] Figure 3 is the heading rejection matrix A in the embodiment of the present invention r The acquisition process. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] Example

[0056] The design idea of ​​the present invention is: to solve the problem of track correlation between radars in radar networking detection scenarios, a dual elimination threshold of speed and heading is used to screen the tracks that require track correlation calculation, which greatly reduces the amount of calculation required for track correlation and improves accuracy.

[0057] See also Figures 1 to 3 The following is a further detailed description of the implementation of the efficient track association method based on the double threshold rejection strategy in conjunction with the accompanying drawings:

[0058] Reference Figure 1 , is a flow chart of an efficient track association method based on a double threshold rejection strategy of the present invention, which is accomplished by including the following steps:

[0059] Step 1: Establish a radar network detection system, and the fusion center receives the track information reported by each radar; complete radar track preprocessing, which at least includes outlier removal and time alignment, and unify the spatial coordinates to the geocentric rectangular coordinate system, and unify parameters such as speed and heading angle at the same time in the same coordinate system.

[0060] In this embodiment, specifically, step one includes the following steps:

[0061] a) Establish a radar network detection system, and the fusion center receives track information reported by various radars;

[0062] b) Complete outlier elimination and eliminate discrete data;

[0063] c) Use interpolation and extrapolation methods to time align radar data;

[0064] d) Perform coordinate conversion to unify the coordinates of all radar-reported tracks into the geocentric rectangular coordinate system, and complete the spatiotemporal unification of parameters such as speed, heading angle, and position.

[0065] Step 2, refer to Figure 2 , obtain the speed information of the two radars R1 and R2, form speed vectors V1 and V2, and calculate the speed screening parameter matrix V according to the speed vectors p , and then the speed screening parameter matrix V p The matrix elements in One by one with the speed error threshold VT u VT d By comparison, the speed in the interval is considered to be in line with the elements, and it can enter the next round of screening. This process is repeated to complete the first round of elimination calculations and obtain the speed elimination matrix V. r ;

[0066] The specific steps are as follows:

[0067] a) First, obtain the speed information of all reported tracks of the two radars R1 and R2 to form speed vectors V1 and V2:

[0068] V1=[v 11 v 12 v 13 …v 1m ];

[0069] V2=[v 21 v 22 v 23 …v 2n ];

[0070] b) Calculate the transposed matrix of V1

[0071]

[0072] c) V2 is multiplied by the m-order identity matrix E to convert it into a square matrix V 2E :

[0073]

[0074] d) Calculate V 2E The inverse matrix

[0075]

[0076] e) Converted into a row vector, V 2v :

[0077]

[0078] f) Calculate the speed screening parameter matrix V p :

[0079]

[0080]

[0081] g) V p Element by element and speed error threshold VT u and VT d Compare and remove the speed matrix V r Assign values ​​to each element of an m×n matrix.

[0082] like:

[0083] or but

[0084] like:

[0085] and but

[0086] Repeat the threshold comparison operation until all matrix elements After the comparison is completed, the velocity elimination matrix V is finally obtained. r .

[0087] Step 3, refer to Figure 3 , remove the matrix V from the above speed r Find the elements whose matrix parameter is 1 Pick The index ij, where i corresponds to the track number of R1 and j corresponds to the track number of R2. The heading angles of the corresponding tracks are taken out to form the heading vectors A1 and A2 of radars R1 and R2. The heading screening parameter matrix A is calculated based on the heading vectors p , and then compared with the heading error threshold AT u , AT dCompare and consider the elements in the interval as matching, and so on, finally complete the heading elimination calculation and get the heading elimination matrix A r ;

[0088] The specific steps are as follows:

[0089] a) First, obtain the heading angles of the two radars R1 and R2 after velocity elimination to form heading vectors A1 and A2:

[0090]

[0091] b) Calculate the transposed matrix of the heading vector A1

[0092]

[0093] c) The heading vector A2 is multiplied by the q-order identity matrix E and converted to a square matrix A 2E :

[0094]

[0095] d) Calculate A 2E The inverse matrix

[0096]

[0097] e) Converted into a row vector, A 2v :

[0098]

[0099] f) Calculate the heading screening parameter matrix A p :

[0100]

[0101]

[0102] g) A p Element by element and heading error threshold AT u and AT d Compare and eliminate the heading matrix A r Assign values ​​to each element of an m×n matrix.

[0103] like:

[0104] or

[0105] but like:

[0106] and

[0107] but Repeat the threshold comparison operation until all elements After the comparison is completed, the heading elimination matrix A is finally obtained. r .

[0108] Step 4: After the above two threshold eliminations, the heading elimination matrix A r The element is 1 The subscript (ij) is called the track candidate mapping group, i corresponds to the track number of R1, and j corresponds to the track number of R2. The double threshold track association algorithm is used to complete the track association, and the amount of calculation at this time will be greatly reduced.

[0109] The specific steps are as follows:

[0110] a) Take out the track of the candidate association mapping (i, j), assuming that the filtering value of radar R1 for the i-th track at time t is And the error covariance of radar R1 for the ith track at time t is At this point, since the time-space calibration has been completed in the previous step 1, it can be calculated directly.

[0111] b) Assume that the state estimation errors of the two radar stations for the same target are statistically independent. And there exist H0 and H1 that satisfy the following conditions:

[0112] H0: and is the track estimate of the same target;

[0113] H2: and It is not the track estimate of the same target;

[0114] c) At time t, for the R estimation error samples from two local nodes, firstly calculate them one by one based on χ 2 The distribution threshold is used for hypothesis testing. If H0 is accepted, the counter is incremented by 1, otherwise the counter remains unchanged. The test statistic using the weighted method is calculated point by point:

[0115]

[0116] Where: α ij (t) is the correlation coefficient between the track i of radar R1 and the track j of radar R2; t is time; is the filtering value of radar R2 for the jth track at time t; is the error covariance of radar R2 for the jth track at time t; T is the matrix transpose calculation.

[0117] When: α ij When (t)≤σ, the counter M increases by 1.

[0118] If: M ≥ L, then it is determined that the radar R1 track i and the radar R2 track j are the observation values ​​of the same track. σ and L are two-step threshold parameters, which are determined according to the actual radar conditions in the test phase. Repeat the above steps to complete the correlation calculation of all tracks between radar R1 and radar R2.

[0119] The present invention receives the track reported by multiple radars through the fusion center. After initialization data preprocessing, the speed information of all the reported tracks of the two radars R1 and R2 is first obtained to form speed vectors V1 and V2; the speed elimination matrix V is calculated. r At the same time, the heading elimination matrix is ​​calculated in parallel, and the relevant track heading angle is taken out to form the heading vectors A1 and A2; after the heading elimination operation again, the heading elimination matrix A is obtained. r . The speed and heading elimination operations can be processed in parallel. After two steps of elimination and screening, the remaining track combinations have been reduced by 70% to 80%. At this time, the track association algorithm is used to complete the track association. The method of the present invention can avoid the use of a point-by-point track distance association algorithm with a large amount of calculation at the beginning, and when the radar detection range is large and the number of tracks is large, it can quickly eliminate the speed and heading mismatched track pairs, greatly reducing the amount of calculation and improving the efficiency of track association.

[0120] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An efficient track association method based on a double threshold rejection strategy, characterized in that: The following steps are involved: Step 1: Establish a radar network detection system, and the fusion center receives track information reported by various radars; Complete radar track preprocessing, which at least includes outlier removal and time alignment, and align the spatial coordinates to the geocentric rectangular coordinate system, and unify the parameters of the speed and heading angle at the same time in the same coordinate system; Step 2: Obtain the speed information of the two radars R1 and R2, form speed vectors V1 and V2, and calculate the speed screening parameter matrix V according to the speed vectors p , and then the speed screening parameter matrix V p The matrix elements in One by one with the speed error threshold VT u VT d Compare, if it is within the interval, it is considered that the speed meets the element, and it can enter the next round of screening, and so on to complete the first round of elimination calculation, and obtain the speed elimination matrix V r ; Step 3: Eliminate the matrix V from the velocity r Find the element whose parameter is 1 Pick The corner mark ij, where i corresponds to the track number of radar R1 and j corresponds to the track number of radar R2, takes out the heading angle of the corresponding track to form the heading vectors A1 and A2 of radars R1 and R2, and calculates the heading screening parameter matrix A according to the heading vectors p , and then compared with the heading error threshold AT u , AT d Compare, if it is within the interval, it is considered that the heading meets the element, and so on, and finally complete the heading elimination calculation to obtain the heading elimination matrix A r ; Step 4: After two threshold eliminations based on steps 2 and 3, the heading elimination matrix A r Elements with parameter 1 The subscript ij is called the track candidate mapping group, where i corresponds to the track number of radar R1 and j corresponds to the track number of radar R2. The double threshold track association algorithm is used to complete the track association. The double-threshold track association algorithm is used to complete track association. The specific steps are as follows: c1) Take out the track of the candidate track mapping group ij, assuming that the filtering value of the radar R1 for the i-th track at time t is And the error covariance of radar R1 for the ith track at time t is c2) Assume that the state estimation errors of the two radar stations for the same target are statistically independent, and there exist H0 and H1 that satisfy the following conditions: H0: and is the track estimate of the same target; H2: and It is not the track estimate of the same target; c3) At time t, for the R estimation error samples from two local nodes, firstly calculate them one by one based on χ 2 The distribution threshold is used for hypothesis testing. If H0 is accepted, the counter is incremented by 1. Otherwise, the counter value remains unchanged. The test statistic using the weighted method is calculated point by point: Among them, α ij (t) is the correlation coefficient between the track i of radar R1 and the track j of radar R2; t is time; is the filtering value of radar R2 for the jth track at time t; is the error covariance of radar R2 for the jth track at time t; T is the matrix transpose calculation; When α ij When (t)≤σ, the counter M is incremented by 1; if: M≥L, it is determined that the radar R1 track i and the radar R2 track j are the observation values ​​of the same track. σ and L are two-step threshold parameters, which are taken according to the actual radar conditions in the test phase. Repeat the above steps to complete the correlation calculation of all tracks between radar R1 and radar R2.

2. The efficient track association method based on the double threshold rejection strategy according to claim 1 is characterized in that: In step 1 shown, the radar tracks are preprocessed, and the inter-radar data are time-aligned using interpolation and extrapolation methods.

3. The efficient track association method based on the double threshold rejection strategy according to claim 1 is characterized in that: In step 2, the velocity elimination matrix V is calculated from the velocity vectors of the two radars R1 and R2. r , the specific steps are as follows: a1) First, obtain the speed information of all reported tracks of radars R1 and R2 to form speed vectors V1 and V2; a2) Calculate the transposed matrix of the velocity vector V1 a3) Multiply the velocity vector V2 by the m-order unit matrix E and convert it into a square matrix V 2E ; a4) Calculate V 2E The inverse matrix a5) Converted into a row vector, V 2v ; a6) Calculate the speed screening parameter matrix V p ,in, a7) The speed screening parameter matrix V p The matrix elements in One by one with the speed error threshold VT u and VT d Compare and remove the speed matrix V r Assign values ​​to each element; like: or but like: and but Repeat the speed error threshold comparison operation until all matrix elements After the comparison is completed, the velocity elimination matrix V is finally obtained. r .

4. The efficient track association method based on the double threshold rejection strategy according to claim 3 is characterized in that In step 3, the heading elimination matrix A is calculated from the heading vectors of radars R1 and R2. r , the specific steps are as follows: b1) First, obtain the heading angles of radars R1 and R2 after velocity elimination to form heading vectors A1 and A2; b2) Calculate the transposed matrix of the heading vector A1 b3) Multiply the heading vector A2 by the q-order unit matrix E and convert it into a square matrix A 2E ; b4) Calculate A 2E The inverse matrix b5) Converted into a row vector, A 2v ; b6) Calculate the heading screening parameter matrix A p ,in, b7), put A p The matrix elements in One by one with the heading error threshold AT u and AT d Compare and eliminate the heading matrix A r Assign values ​​to each element; like: or but like: and but Repeat the heading error threshold comparison operation until all matrix elements After the comparison is completed, the heading elimination matrix A is finally obtained. r .

Images