A track association method based on multi-factor and multi-model selection

Abstract

The present invention provides a track association method based on multi-factor multi-model selection, comprising the following steps: S10, selecting a fuzzy factor set and an evaluation set. S20, calculating a weight fuzzy set based on the fuzzy factor set and the evaluation set. S30, calculating a correlation degree fuzzy set based on the fuzzy factor set, the evaluation set and the weight fuzzy set. S40, making an association decision for the track based on the correlation degree fuzzy set. The present invention overcomes the defect of the existing method that the same factor and synthetic operation model cannot be applied to all evaluation levels, and improves the association performance.

Description

Technical Field

[0001] The present invention relates to the technical field of multi-source sensor track association and fusion, and in particular to a track association method based on multi-factor and multi-model selection. Background Art

[0002] With the development of modern information technology, the number of sensor types is increasing, and multi-source information is becoming increasingly complex. Multi-source sensor fusion technology has attracted increasing attention from researchers. Sensor information fusion mainly includes target detection, recognition, association, fusion, and state estimation. Among them, target association is the foundation of fusion and determines the upper limit of comprehensive recognition.

[0003] Distributed systems are the most economical and reasonable choice for engineering applications, achieving high reliability and availability at a low cost. In track fusion distributed systems, the key to fusion performance is the correct, stable, and efficient association of tracks from different sensors. Multi-sensor data association in a distributed system is track-to-track association, the purpose of which is to determine whether tracks from different sensors belong to the same target. Track association is a prerequisite for track fusion and determines the upper limit of track fusion performance. Correct association is the basis for stable target tracking. Therefore, track association is the key to multi-sensor information fusion, and studying track association is necessary and meaningful.

[0004] The purpose of association is to determine whether the measurements of different sensors belong to the same target. Track association algorithms are divided into statistical-based track association algorithms and fuzzy mathematics-based track association algorithms according to whether they are statistical test problems. Track association methods based on statistics include the nearest neighbor method, weighted method, sequential method, multi-dimensional allocation method, etc. Methods based on fuzzy mathematics include the fuzzy comprehensive function method, multi-factor fuzzy decision-making method, etc. The statistical-based method converts the track association problem into a hypothesis testing problem. Due to navigation, sensor delay and error problems, the association has a large ambiguity. Therefore, the statistical-based method cannot achieve good association performance in scenarios such as dense targets. The fuzzy mathematics-based method introduces association ambiguity to describe the correlation between tracks. It has the ability to handle uncertainty and is suitable for association scenarios with large fuzzy relationships.

[0005] Traditional multi-factor fuzzy decision-making methods use the same set of factors and synthesis model to generate the corresponding membership degrees when calculating fuzzy sets with different degrees of association. However, it is worth noting that different evaluations use different factors and synthesis models. Using the same factors and synthesis model for all evaluation levels will lead to a decrease in the accuracy of the association membership degrees, thereby reducing the association performance. This is specifically reflected in two aspects:

[0006] First, different factors are evaluated differently. If the same factors are used for correlation calculations across all evaluation levels, correlation performance will degrade. This is particularly true in environments with dense air-surface clutter and multiple neighbors (intersecting or bifurcating), where multiple targets are moving at the same speed, high similarity between their speeds will affect (or increase) the overall track similarity. Similarly, when multiple targets are moving at the same heading, similarity between their headings will affect the overall track correlation quality. Therefore, using speed and heading as fuzzy factors to measure target similarity will degrade correlation performance.

[0007] Second, different evaluation criteria utilize different synthetic computational models. "Association" considers the overall effect of the similarity of various factors, while "unassociation" considers the fuzzy factor that contributes the greatest dissimilarity. Therefore, using the same synthetic model for both "association" and "unassociation" will reduce the expressive power of the association evaluation and degrade association performance. Summary of the Invention

[0008] The present invention aims to provide a track association method based on multi-factor and multi-model selection to solve the problem that the same factors and synthetic operation models used in the existing multi-factor fuzzy comprehensive decision-making method cannot be applied to all evaluation levels.

[0009] The present invention provides a track association method based on multi-factor and multi-model selection, comprising the following steps:

[0010] S10, select the fuzzy factor set and evaluation set.

[0011] S20, calculating a weighted fuzzy set based on the fuzzy factor set and the evaluation set.

[0012] S30, calculating the association degree fuzzy set based on the fuzzy factor set, the evaluation set and the weight fuzzy set.

[0013] S40, making association decisions on the tracks based on the association degree fuzzy set.

[0014] Furthermore, step S10 includes the following sub-steps:

[0015] S11, select the fuzzy factor set: the Euclidean distance between the two target positions, the Euclidean distance between the two track target velocities, and the Euclidean distance between the two track target headings are used as fuzzy factors, a total of 3 fuzzy factors, and the fuzzy factor set is expressed as:

[0016] U={u1,u2,u3}

[0017] Among them, u1 represents the Euclidean distance between the two target positions, u2 represents the Euclidean distance between the two track target velocities, and u3 represents the Euclidean distance between the two track target headings;

[0018] S12, select the membership function of the evaluation set:

[0019] Taking the normal membership as a measure of the similarity between two tracks, the membership of the kth fuzzy factor is expressed as:

[0020]

[0021] Among them, σ k represents the spread of the kth fuzzy factor, u k Indicates the adjustment degree of the kth fuzzy factor.

[0022] When judging by two evaluation levels, there are k1 +r k2 =1, therefore, the non-membership degree of the kth fuzzy factor that measures the dissimilarity between two trajectories is expressed as:

[0023]

[0024] S13, select evaluation set V = {related, not related}.

[0025] Furthermore, step S20 includes the following sub-steps:

[0026] S21, the weighted fuzzy set is expressed as:

[0027]

[0028] Where A1=[a 11 ,a 12 ,a 13 ],A2=[a 21 ,a 22 ,a 23 ], where A1 represents the weighted fuzzy set used to measure the “association” evaluation, A2 represents the weighted fuzzy set used to measure the “non-association” evaluation, and a ij It represents the weight value when using the jth fuzzy factor to measure the i-th evaluation;

[0029] S22, sufficiency and necessity measurement of fuzzy factors: analyzing the correlation characteristics of each fuzzy factor, specifically analyzing the sufficiency and necessity between each fuzzy factor and the correlation evaluation;

[0030] S23, based on the sufficiency and necessity measure of fuzzy factors, the weight of each fuzzy factor under each associated evaluation is derived, thereby obtaining a weighted fuzzy set.

[0031] Furthermore, in step S22, the sufficiency and necessity between each fuzzy factor and the associated evaluation are as follows:

[0032] If the similarity of the fuzzy factors between two tracks is high enough to deduce the association, then the sufficiency of the fuzzy factors and the “association” is satisfied; if the association between the tracks can deduce the high similarity of the fuzzy factors, then the necessity of the fuzzy factors and the “association” is satisfied;

[0033] If the similarity of the fuzzy factors between two tracks is low and it can be deduced that they are not correlated, then the sufficiency of the fuzzy factor and "not correlated" is satisfied. If the non-correlation between the tracks can be deduced that the similarity of the fuzzy factors is low, then the necessity of the fuzzy factor and "not correlated" is satisfied.

[0034] Furthermore, step S23 includes:

[0035] If the fuzzy factor and the correlation evaluation do not meet the sufficiency, the corresponding weight is 0, and the weight fuzzy set A1 used to measure the "correlation" evaluation is expressed as A1 = [a 11 ,0,0]=[1,0,0], the weighted fuzzy set A2 used to measure the “non-correlated” evaluation is expressed as A2=[a 21 ,a 22 ,a 23 ]=[0.33,0.33,0.33].

[0036] Furthermore, step S30 includes the following sub-steps:

[0037] S31, constructing a multivariate synthesis model based on the fuzzy factor set and the evaluation set, wherein the multivariate synthesis model selects different synthesis operation models according to the characteristics of different evaluation levels;

[0038] S32, calculating the association degree fuzzy set based on the multivariate synthesis model and the weighted fuzzy set.

[0039] Furthermore, step S31 includes:

[0040] When evaluating “relevance”, it is the joint influence of each fuzzy factor that affects the relevance judgment;

[0041] When evaluating “non-correlation”, the fuzzy factor that has the greatest impact on the non-correlation judgment is selected.

[0042] Furthermore, in step S32, the fuzzy set of association degree is calculated Expressed as:

[0043]

[0044]

[0045] Among them, R represents the membership function of the evaluation set.

[0046] Furthermore, step S40 includes the following sub-steps:

[0047] S41, based on the fuzzy set of association degree, the closeness matrix G(l) is constructed, which is expressed as:

[0048]

[0049] Among them, n A Represents the number of tracks of sensor A, n B represents the number of tracks of sensor B, b ij1 (l) and b ij2 (l) are b1 and b2 of track i and track j at time l respectively;

[0050] S42, track association judgment, if g ij *(l)>δ, then it is determined that track i is associated with track j* at time l, δ is the association threshold, and at the same time, U A Represents the track of sensor A, let g i'j* (l) = 0, and then repeat the above formula to make track association judgment.

[0051] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0052] This invention utilizes the correlation characteristics of various factors and the applicability of the synthetic operation model in different evaluations, calculates the correlation degree fuzzy set through multi-factor directional selection and multivariate synthetic operation model, and generates track correlation results. It is mainly manifested in two aspects:

[0053] First, based on the analysis of the sufficiency and necessity between different factors and different evaluations, different factors are measured with different correlation evaluations, which effectively combines the correlation characteristics of different factors and improves the correlation performance;

[0054] Second, a multivariate synthesis model is proposed. This multivariate synthesis model selects the corresponding model according to the correlation evaluation properties, quantitatively reflecting the degree of correlation and non-correlation between two tracks. Compared with a single synthesis model, it more accurately reflects the characteristics of the evaluation level.

[0055] In summary, compared with the existing multi-factor fuzzy comprehensive decision-making algorithm, the present invention overcomes the defect of the existing method that the same factor and synthetic operation model cannot be applied to all evaluation levels, and improves the correlation performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings in the embodiments will be briefly introduced below. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0057] Figure 1 Flowchart of a track association method based on multi-factor and multi-model selection in an embodiment of the present invention.

[0058] Figure 2 Schematic diagram of the logical relationship between fuzzy factors and associated evaluation in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of 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. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0060] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0061] Example

[0062] like Figure 1 As shown, this embodiment proposes a track association method based on multi-factor and multi-model selection, including the following steps:

[0063] S10, select the fuzzy factor set and evaluation set. When making multi-factor fuzzy decisions, the fuzzy factors and evaluation criteria in the track must be analyzed first. Only after the fuzzy factor set and evaluation set are obtained can subsequent processing be carried out. The specific sub-steps include the following:

[0064] S11, select the fuzzy factor set, and use the Euclidean distance between the two target positions, the Euclidean distance between the two track target velocities, and the Euclidean distance between the two track target headings as fuzzy factors, a total of 3 fuzzy factors. The fuzzy factor set is expressed as:

[0065] U={u1,u2,u3}

[0066] Among them, u1 represents the Euclidean distance between the two target positions, u2 represents the Euclidean distance between the two track target velocities, and u3 represents the Euclidean distance between the two track target headings.

[0067] Assumptions is the state estimation vector of track i in sensor A, is the state estimation vector of track j in sensor B, and the three fuzzy factors are expressed as:

[0068]

[0069]

[0070]

[0071] S12, select the membership function of the evaluation set:

[0072] Taking the normal membership as a measure of the similarity between two tracks, the membership of the kth fuzzy factor is expressed as:

[0073]

[0074] Among them, σ k represents the spread of the kth fuzzy factor, u k Indicates the adjustment degree of the kth fuzzy factor.

[0075] When judging by two evaluation levels, there are k1 +r k2 =1, therefore, the non-membership degree of the kth fuzzy factor that measures the dissimilarity between two trajectories is expressed as:

[0076]

[0077] S13, select the evaluation level, and set the evaluation set V = {v1, v2, ..., v m} in which m is selected as 2, that is:

[0078] V = {association, non-association}

[0079] S20, calculates a weighted fuzzy set based on the fuzzy factor set and the evaluation set. Weighted fuzzy sets are the key to solving problems using fuzzy theory. By analyzing the correlation characteristics of each fuzzy factor, the weight of each fuzzy factor under each correlation evaluation is derived. This specifically includes the following sub-steps:

[0080] S21, the weighted fuzzy set is expressed as:

[0081]

[0082] Where A1=[a 11 ,a 12 ,a 13 ],A2=[a 21 ,a 22 ,a 23 ], where A1 represents the weighted fuzzy set used to measure the “association” evaluation, A2 represents the weighted fuzzy set used to measure the “non-association” evaluation, and a ij It represents the weight value when using the jth fuzzy factor to measure the i-th evaluation.

[0083] S22, sufficiency and necessity measurement of fuzzy factors. Analyze the correlation characteristics of each fuzzy factor, specifically analyze the sufficiency and necessity between each fuzzy factor and the correlation evaluation, such as Figure 2 As shown:

[0084] If the similarity of the fuzzy factors between two tracks is high enough to deduce the association, then the sufficiency of the fuzzy factors and the “association” is satisfied; if the association between the tracks can deduce the high similarity of the fuzzy factors, then the necessity of the fuzzy factors and the “association” is satisfied;

[0085] Similarly, if the similarity of the fuzzy factors between two tracks is low and it can be deduced that they are not correlated, then the sufficiency of the fuzzy factor and “not correlated” is satisfied; if the non-correlation between the tracks can be deduced that the similarity of the fuzzy factors is low, then the necessity of the fuzzy factor and “not correlated” is satisfied.

[0086] Therefore, the relationship between the fuzzy factors "position" and "correlation" satisfies the sufficiency and necessity, the relationship between "speed" and "course" and "correlation" only satisfies the necessity; the relationship between "position", "speed", "course" and "no correlation" all satisfies the sufficiency and necessity.

[0087] If the fuzzy factors and associations meet the necessary and sufficient conditions, then:

[0088] Establish, i∈U1, j∈U2

[0089] Establish, i∈U1, j∈U2

[0090] Among them, H0 means that track i and track j are the same target, and H1 means that track i and track j are different targets.

[0091] S23, according to the sufficiency and necessity measure of fuzzy factors, derive the weight of each fuzzy factor under each correlation evaluation: if the fuzzy factor and the correlation evaluation do not meet the sufficiency, the corresponding weight is 0, and the weight fuzzy set A1 used to measure the "correlation" evaluation is expressed as A1 = [a 11 ,0,0]=[1,0,0], the weighted fuzzy set A2 used to measure the “non-correlated” evaluation is expressed as A2=[a 21 ,a 22 ,a 23 ]=[0.33,0.33,0.33].

[0092] S30, calculating the association degree fuzzy set based on the fuzzy factor set, the evaluation set, and the weighted fuzzy set. The association degree fuzzy set measures the degree of membership of the evaluation set and is the joint result of the weighted fuzzy set and the fuzzy relationship matrix through a multivariate synthesis model. Specifically, it includes the following sub-steps:

[0093] S31. Construct a multivariate synthesis model based on the fuzzy factor set and the evaluation set. This multivariate synthesis model selects different synthesis operation models based on the characteristics of different evaluation levels. When evaluating "association," the combined influence of each fuzzy factor influencing the association judgment is used; when evaluating "non-association," the fuzzy factor with the greatest influence on the non-association judgment is selected. Thus, the constructed multivariate synthesis model consists of a weighted average model and a main factor prominence model, suitable for different association evaluations.

[0094] S32, Calculate the degree of association fuzzy set based on multivariate synthesis model and weighted fuzzy set Expressed as:

[0095]

[0096]

[0097] Among them, R represents the membership function of the evaluation set.

[0098] S40, performing association decision on the tracks based on the association degree fuzzy set. The association decision ultimately determines whether two tracks are generated by the same target. After calculating the closeness matrix using the association degree fuzzy set, a threshold is set to determine whether they are associated. This specifically includes the following sub-steps:

[0099] S41, based on the fuzzy set of association degree, the closeness matrix G(l) is constructed, which is expressed as:

[0100]

[0101] Among them, n A Represents the number of tracks of sensor A, n B represents the number of tracks of sensor B, b ij1 (l) and b ij2 (l) are b1 and b2 of track i and track j at time l respectively.

[0102] S42, track association judgment, if g ij *(l)>δ, then it is determined that track i is associated with track j* at time l, δ is the association threshold, and at the same time, U A Represents the track of sensor A, let g i'j* (l) = 0, and then repeat the above formula to make track association judgment.

[0103] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A track association method based on multi-factor and multi-model selection, characterized in that: The steps include: S10, select the fuzzy factor set and evaluation set; S20, calculating the weighted fuzzy set based on the fuzzy factor set and the evaluation set; S30, calculating the association degree fuzzy set based on the fuzzy factor set, the evaluation set and the weight fuzzy set; S40, making association decisions on the tracks based on the fuzzy set of association degree; Step S10 includes the following sub-steps: S11, select a fuzzy factor set: the Euclidean distance between the two target positions, the Euclidean distance between the two track target speeds, and the Euclidean distance between the two track target headings are used as fuzzy factors, a total of 3 fuzzy factors, and the fuzzy factor set is expressed as: U={u1,u2,u3} Among them, u1 represents the Euclidean distance between the two target positions, u2 represents the Euclidean distance between the two track target velocities, and u3 represents the Euclidean distance between the two track target headings; S12, select the membership function of the evaluation set: Taking the normal membership as a measure of the similarity between two tracks, the membership of the kth fuzzy factor is expressed as: Among them, σ k represents the spread of the kth fuzzy factor, u k represents the adjustment degree of the kth fuzzy factor; When judging by two evaluation levels, there are r k1 +r k2 =1, therefore, the non-membership degree of the kth fuzzy factor that measures the dissimilarity of the two trajectories is expressed as: S13, select evaluation set V = {related, not related}; Step S20 includes the following sub-steps: S21, the weight fuzzy set is expressed as: Where A1 = [a 11 ,a 12 ,a 13 ],A2=[a 21 ,a 22 ,a 23 ], where A1 represents the weighted fuzzy set used to measure the "association" evaluation, A2 represents the weighted fuzzy set used to measure the "non-association" evaluation, and a ij It represents the weight value when the jth fuzzy factor is used to measure the i-th evaluation; S22, sufficiency and necessity measurement of fuzzy factors: analyzing the correlation characteristics of each fuzzy factor, specifically analyzing the sufficiency and necessity between each fuzzy factor and the correlation evaluation; S23, according to the necessary and sufficient measure of fuzzy factors, the weight of each fuzzy factor under each associated evaluation is derived, thus obtaining the weighted fuzzy set; In step S22, the necessity and sufficiency between each fuzzy factor and the associated evaluation is as follows: If the similarity of the fuzzy factors between two tracks is high and the association can be derived, then the sufficiency of the fuzzy factor and the "association" is satisfied; if the association between the tracks can deduce the high similarity of the fuzzy factors, then the necessity of the fuzzy factor and the "association" is satisfied; If the similarity of the fuzzy factors between two tracks is low and can be deduced to be uncorrelated, then the sufficiency of the fuzzy factor and "uncorrelated" is satisfied; if the uncorrelated between the tracks can be deduced to have a low similarity of the fuzzy factors, then the necessity of the fuzzy factor and "uncorrelated" is satisfied; Step S30 includes the following sub-steps: S31, constructing a multivariate synthesis model based on the fuzzy factor set and the evaluation set, wherein the multivariate synthesis model selects different synthesis operation models according to the characteristics of different evaluation levels; S32, calculating the fuzzy set of association degree based on the multivariate synthesis model and the weighted fuzzy set; Step S31 includes: When evaluating "relevance", it is the combined effect of each fuzzy factor that affects the relevance judgment; When evaluating "irrelevance", the fuzzy factor that has the greatest impact on the irrelevance judgment is selected; Step S40 includes the following sub-steps: S41, based on the fuzzy set of association degree, the closeness matrix G(l) is constructed, which is expressed as: Among them, n A Represents the number of tracks of sensor A, n B represents the number of tracks of sensor B, b ij1 (l) and b ij2 (l) are b1 and b2 of track i and track j at time l respectively; S42, track association decision, if g ij* (l)>δ, then it is determined that track i is associated with track j* at time l, δ is the association threshold, and at the same time, U A represents the track of sensor A, let g i'j* (l) = 0, and then repeat the above formula to make track association judgment.

2. The track association method based on multi-factor and multi-model selection according to claim 1 is characterized in that: Step S23 includes: If the fuzzy factor and the correlation evaluation do not meet the sufficiency, the corresponding weight is 0. The weight fuzzy set A1 used to measure the "correlation" evaluation is expressed as A1 = [a 11 ,0,0]=[1,0,0], the weighted fuzzy set A2 used to measure the "irrelevant" evaluation is expressed as A2=[a 21 ,a 22 ,a 23 ]=[0.33,0.33,0.33].

3. The track association method based on multi-factor multi-model selection according to claim 2 is characterized in that: In step S32, the fuzzy set of association degree is calculated It is expressed as: Among them, R represents the membership function of the evaluation set.

Images