Motion entity correlation method based on grey correlation matrix

Abstract

The invention discloses a motion entity association method based on a grey incidence matrix, and belongs to the technical field of artificial intelligence and knowledge engineering. According to the method, an entity association judgment method combining attribute characteristics and state information of a motion entity is adopted, state information such as the position, the speed, the acceleration and the azimuth angle of the entity and related attribute characteristics serve as indexes to form an association matrix, and the association degree between matrixes at adjacent moments is calculated and compared to determine the association relation of the entity. The method can be used for solving the problem of entity association of multi-source heterogeneous information, and provides support and basis for the real-time fusion technology of moving entities.

Description

Technical Field

[0001] The present invention relates to the technical fields of artificial intelligence and knowledge engineering, and particularly relates to a motion entity association method based on a grey correlation matrix. Background Art

[0002] In modern conflicts, the multi-source motion entity data association technology obtained by sensors of different systems is the core technology for realizing the functions of motion entity processing and attribute fusion recognition. Traditional association algorithms such as the joint probabilistic data association algorithm have good effects on multi-radar data association processing. However, for motion entity-related data such as AIS data, technical detection data, and movement trend data, due to differences in accuracy, time delay, etc., the original association algorithms cannot effectively solve the association problem between multi-source heterogeneous data. Therefore, it is urgent to study an entity association method that can handle motion entity-related data including radar data, AIS data, technical detection data, movement trend data, etc. Summary of the Invention

[0003] In view of this, the present invention provides a motion entity association method based on a grey correlation matrix. By an entity association judgment method that combines the attribute characteristics and state information of motion entities, the state information such as the position, speed, acceleration, and azimuth angle of the entity and relevant attribute characteristics are used as indicators to form a correlation matrix, and the correlation degree between adjacent time matrixes is calculated and compared to determine the entity association relationship.

[0004] The present invention can be achieved by the following technical means:

[0005] A motion entity association method based on a grey correlation matrix, comprising the following steps:

[0006] Step 1, setting characteristic indicators for associating entities, including motion state indicators and attribute indicators;

[0007] Step 2, constructing an entity library, assigning real values to each characteristic indicator of each entity, and each entity is uniquely determined by all its characteristic indicators;

[0008] Step 3, in a real scenario, according to the observed attribute indicators of physical objects, perform rough association in the entity library, find the entities that match the attribute indicators of the physical objects, and obtain a rough association entity library X=(x 1 , x 2 , …, x m ), there are m entities in the rough association entity library, each entity has n motion state indicators, and the jth motion state indicator of the ith entity is x ij ;

[0009] Step 4, converting each motion state indicator of each entity in the rough association entity library into a dimensionless value;

[0010] Step 5: Collect various motion state indicators of the physical object, convert each motion indicator into a dimensionless value, and use the grey relational analysis method to calculate the grey relational degree between the physical object and each entity in the rough correlation entity library;

[0011] Step 6: Sort the grey relational degrees between the physical object and each entity in the rough correlation entity library from small to large, and the entity with the largest grey relational degree is the correlation result of the physical object.

[0012] Furthermore, there are 7 motion state indicators in total, namely: longitude position, latitude position, meridional velocity, zonal velocity, meridional acceleration, zonal acceleration, and motion direction. Among them, the motion direction is expressed as the angle with the due north direction; the attribute indicators include: name, type, model, and activity area.

[0013] Furthermore, in Step 4, the mean normalization method is used to perform dimensionless calculation on the motion state indicators. The specific method is as follows:

[0014]

[0015] where is the dimensionless value of the motion state indicator x ij after conversion.

[0016] Furthermore, in Step 5, the calculation method of the grey relational degree is as follows:

[0017] Step 501: Collect various motion state indicators of the physical object, denoted as x 0 =(x 01 , x 02 , …, x 0n ), where x 0n represents the nth motion state indicator of the physical object;

[0018] Step 502: Use the mean normalization method to perform dimensionless calculation on the motion state indicators of the collected physical object. The specific method is as follows:

[0019]

[0020] where is the dimensionless value of the motion state indicator x 0j after conversion;

[0021] Step 503: Calculate the grey relational coefficient between the jth motion state indicator x 0j of the physical object and the jth motion state indicator x ij of the ith entity in the rough correlation entity library:

[0022]

[0023] Among them, ρ is the resolution coefficient, where 0 < ρ < 1, and Δ ij is the absolute difference between the corresponding value of the physical object and the entity object in the rough association entity library, and Δ min is the two-level minimum difference, and Δ max is the two-level maximum difference, and the calculation method is as follows:

[0024]

[0025] Step 504, calculate the gray correlation degree r 0 between the physical object x i and the i-th entity x oi in the rough association entity library:

[0026]

[0027] Among them, W j is the weight of the gray correlation coefficient r oi , W 1 + W 2 + … + W n = 1.

[0028] The beneficial effects of the present invention are as follows:

[0029] 1. The present invention adopts an entity association judgment method that combines the attribute characteristics and state information of moving entities, forms an association matrix with state information such as the position, speed, acceleration, and azimuth angle of the entity and relevant attribute characteristics as indicators, and determines the entity association relationship by calculating and comparing the association degrees between matrices at adjacent moments.

[0030] 2. The method of the present invention can be used to solve the entity association of multi-source heterogeneous information, providing support and basis for the real-time fusion technology of moving entities. Specific embodiments

[0031] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below.

[0032] A method for associating moving entities based on a gray correlation matrix includes the following steps:

[0033] Step 1, set the characteristic indicators for associating entities, and use the longitude position, latitude position, meridional velocity, zonal velocity, meridional acceleration, zonal acceleration, and movement direction of the entity as the motion state indicators; use the name, type, model, and activity area of the entity as the attribute indicators;

[0034] Velocity characterizes the speed of entity movement; acceleration characterizes the maneuverability of the entity; the entity direction characterizes the action trend of the entity. Therefore, position, velocity, acceleration, and the entity direction are used as the state indicators for association.

[0035] The entity type represents the classification of the entity, such as aircraft, ship, etc.; the equipment model represents the specific model of the entity; the entity name represents the specific identity of the entity; the activity area represents the activity range of the entity. Therefore, the entity type, model, name, and activity area are used as attribute indicators.

[0036] Step 2, construct an entity library, assign real values to each characteristic index of each entity, and each entity is uniquely determined by all its characteristic indexes;

[0037] Step 3, in the real scenario, according to the observed attribute indicators of the physical object, conduct a rough association in the entity library to find the entity that matches the attribute indicators of the physical object, and obtain the rough association entity library X = (x 1 , x 2 , …, x m ), there are m entities in the rough association entity library, each entity has n motion state indicators, and the jth motion state indicator of the ith entity is x ij ;

[0038] Step 4, adopt a dimensionless processing formula to conduct dimensionless data processing on the entity characteristic indexes;

[0039] The physical meanings of each index are different, resulting in different dimensions of the data, which is not convenient for comparison or difficult to obtain correct conclusions when comparing. Therefore, when conducting grey relational analysis, it is necessary to conduct dimensionless data processing first. Suppose there are m entities with successful rough correlation in the entity library, denoted as X = (x 1 , x 2 , …, x m ), each entity has n characteristic indexes, and x ij is the jth index in the ith entity. The mean method is used for dimensionless data processing, and the formula is as follows:

[0040]

[0041] Among them, is the dimensionless value of the converted motion state indicator x ij .

[0042] Step 5, collect each motion state index of the physical object, convert each motion index into a dimensionless value, and use the grey relational analysis method to calculate the grey relational degree between the physical object and each entity in the rough association entity library;

[0043] The grey relational degree describes the relative change situation between system factors. If the relative changes are basically the same, it is considered that the relational degree between the two is large; otherwise, the relational degree between the two is small.

[0044] In the grey relational analysis theory, the rough relational entity x in the entity library i =(x i1 , x i2 , …, x in ), i = 1, 2, …, m is called the comparison sequence, and the current observation of the sensor is x 0 =(x 01 , x 02 , …, x 0n ), which is called the reference sequence. The grey relational coefficient of the j-th index of the reference sequence x 0 and the comparison sequence x i is as follows:

[0045]

[0046] where Δ min is called the two-level minimum difference, Δ max is the two-level maximum difference, and Δ ij is the absolute difference between the corresponding values of the physical object and the entity object in the rough relational entity library, which is defined as follows:

[0047]

[0048] ρ< is the resolution coefficient, and generally ρ = 0.5. reflects the degree of association between the current observation and the j-th characteristic attribute of the i-th entity.

[0049] The calculation method of the grey relational degree is as follows:

[0050]

[0051] where the grey relational degree r oi is a measure of the geometric distance between the reference sequence x 0 and the comparison sequence x i , W j is the weight of the grey relational coefficient r oi , and W 1 +W 2 +…+W n = 1.

[0052] Step 6, according to the resolution principle of the grey system, sort the grey relational degrees. If it satisfies:

[0053]

[0054] then it is judged that the current observation x 0 is associated with the entity x k .

[0055] In summary, the present invention adopts an entity association judgment method that combines the attribute characteristics of moving entities with status information, forms an association matrix using status information such as the position, speed, acceleration, and azimuth angle of entities and relevant attribute characteristics as indicators, and determines the entity association relationship by calculating and comparing the association degrees between matrices at adjacent moments. The method of the present invention can be used to solve the entity association of multi-source heterogeneous information and provide support and basis for the real-time fusion technology of moving entities.

Claims

1. A motion entity association method based on grey association matrix, characterized in that: The steps include: Step 1, setting feature indicators for associating entities, including motion state indicators and attribute indicators; Step 2: Build an entity library and assign real values ​​to each characteristic index of each entity. Each entity is uniquely identified by all its characteristic indexes. Step 3: In the real scene, according to the attribute indicators of the observed physical objects, a rough association is performed in the entity library to find the entities that match the attribute indicators of the physical objects, and a rough association entity library X = (x1, x2, ..., x m ), there are m entities in the coarse association entity library, each entity has n motion state indicators, and the jth motion state indicator of the i-th entity is x ij ; Step 4, converting each motion state index of each entity in the coarse-association entity library into a dimensionless value; Step 5, collecting various motion state indicators of the physical object, converting each motion indicator into a dimensionless value, and using a grey correlation analysis method to calculate the grey correlation degree between the physical object and each entity in the rough correlation entity library; Step 6: Sort the grey correlation between the physical object and each entity in the coarse correlation entity library from small to large, and the entity with the largest grey correlation is the correlation result of the physical object.

2. A method for associating moving entities based on grey association matrix according to claim 1, characterized in that: There are seven motion status indicators, namely: longitude position, latitude position, longitudinal speed, latitudinal speed, longitudinal acceleration, latitudinal acceleration, and motion direction, where the motion direction is expressed as the angle with the north direction; the attribute indicators include: name, type, model, and activity area.

3. The motion entity association method based on grey association matrix according to claim 1 is characterized in that: In step 4, the averaging method is used to perform dimensionless calculation on the motion state index, specifically: in, is the converted motion state index x ij The dimensionless value of .

4. The motion entity association method based on grey association matrix according to claim 1 is characterized in that: In step 5, the calculation method of grey relational degree is: Step 501, collect various motion state indicators of the physical object, expressed as x0=(x 01 ,x 02 ,…,x 0n ), where x 0n Represents the nth motion state indicator of the physical object; Step 502, using the averaging method to perform dimensionless calculation on the motion state index of the collected physical object, the specific method is: in, is the converted motion state index x 0j The dimensionless value of Step 503: Calculate the j-th motion state index x of the physical object 0j The jth motion state index x of the i-th entity in the coarsely associated entity library ij Grayscale correlation coefficient: Where ρ is the resolution coefficient, 0<ρ<1, Δ ij is the absolute difference between the physical object and the corresponding value of the entity object in the coarse-associated entity library, Δ min is the minimum difference between the two levels, Δ max The maximum difference between the two levels is calculated as follows: Step 504: Calculate the relationship between the physical object x0 and the i-th entity x in the coarsely associated entity library. i The gray correlation r oi : Among them, W j is the grayscale correlation coefficient r oi The weight of W1+W2+…+W n =1.

Images