A multi-source detection fusion algorithm for a multi-layered interception and defense system

Abstract

This invention relates to a multi-source detection fusion algorithm for a multi-layered interception and defense system, comprising the following steps: S1, detecting moving targets using multi-source detectors to acquire different feature information; S2, obtaining an information matrix acquired by multiple detection methods; classifying the sampled data at each sampling time according to feature type, dividing the detection data to be analyzed into m categories; selecting data with similar features for mapping and generalization processing to obtain a detection information function, then extracting the required feature types to obtain a feature objective function; and constructing the optimal objective function to obtain the global optimal solution. The multi-source detection fusion algorithm for a multi-layered interception and defense system provided by this invention can fuse detection information from various detectors, perform more diverse feature fusion selection, and, for the same features, more closely approximate the optimal detection value, thus making target detection and identification easier.

Description

Technical Field

[0001] This invention belongs to the field of defense detection and surveillance technology, and in particular relates to a multi-source detection fusion algorithm for a multi-layered interception and defense system. Technical Background

[0002] In recent years, the use of low, slow and small aircraft for intelligence reconnaissance has emerged. Low, slow and small aircraft refer to light drones, model aircraft, multi-rotor drones, fixed-wing drones, and unmanned helicopters. These low, slow and small aircraft are generally difficult to detect and have short warning times. For these low, slow and small swarm aircraft, the existing defense system has relatively simple detection methods and poor ability to distinguish them.

[0003] Existing information fusion technologies for low, small, and slow targets in the air are widely used, but most fusion algorithms fuse detection information from similar detectors, making it difficult to combine the advantages of various detectors to obtain the optimal detection results. In order to calculate target feature information more accurately, a new target information fusion algorithm is needed to fuse feature information from different detectors in different dimensions to obtain the optimal fusion solution. Summary of the Invention

[0004] To address the aforementioned issues, this invention proposes a multi-source detection fusion algorithm for a multi-layered interception and defense system. This algorithm can fuse target information from different sources and detection types in real time and perform target feature classification. This invention combines the advantages of different detectors, fuses feature information from different dimensions, obtains the optimal algorithm fusion solution, and achieves real-time target defense.

[0005] The technical solution adopted in this invention is a multi-source detection fusion algorithm for a multi-layered interception and defense system, including the following steps:

[0006] S1. Detect moving targets using multi-source detectors to obtain different feature information;

[0007] Detection methods include Doppler radar detection, radar imaging detection, optical detection, and infrared detection;

[0008] S2. Perform multi-objective information fusion processing, which includes the following steps:

[0009] S21, Target Differentiation Algorithm;

[0010] U1 = [a1, a2, a3, a4, ... an];

[0011] U2 = [b1, b2, b3, b4, ... bn];

[0012] U3 = [c1,c2,c3,c4,...cn];

[0013] U4 = [d1, d2, d3, d4, ... dn];

[0014] X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn,d1,d2,d3,d4,...dn]

[0015] U1, U2, U3, and U4 are the information matrices obtained from the four detection methods, respectively.

[0016] X is a matrix containing all the information obtained from the four detection methods;

[0017] S22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n.

[0018] S23. Select data with similar characteristics and perform mapping generalization processing to obtain the detection information function V=[Vij|i=1,2...,n,j=1,2,...,s], and then extract the required feature types;

[0019] S24. Obtain the characteristic objective function;

[0020]

[0021] In the formula, m is the category dimension of the feature, and μ ik For the weighting of each feature, d ik denoted as Euclidean distance between feature nodes and feature centers, where c is the number of meaningful feature categories extracted from m categories, 1 < c < m;

[0022] (d ik ) 2 =||x k -V i || 2 , where X k The feature node is the k-th item in the X matrix; v i The central node is initialized with an assigned empirical value, and the optimal solution obtained from each subsequent calculation is used for iteration.

[0023] and

[0024] S25. Construct the optimal objective function:

[0025] S26. Obtain the global optimal solution.

[0026] This invention also provides a multi-source detection fusion algorithm for a multi-layered interception and defense system, characterized by comprising the following steps:

[0027] Q1. Detect moving targets using multi-source detectors to obtain different feature information;

[0028] Detection methods include Doppler radar detection, radar imaging detection, and optical detection;

[0029] Q2. Perform multi-objective information fusion processing, which includes the following steps:

[0030] Q21. Target discrimination algorithm;

[0031] U1 = [a1, a2, a3, a4, ... an];

[0032] U2 = [b1, b2, b3, b4, ... bn];

[0033] U3 = [c1,c2,c3,c4,...cn];

[0034] X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn]

[0035] U1, U2, and U3 are the information matrices obtained by the three detection methods, respectively.

[0036] X is a matrix containing all the information obtained from the three detection methods;

[0037] Q22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n.

[0038] Q23. Select data with similar characteristics and perform mapping generalization to obtain the detection information function V = [Vij|i = 1, 2, ..., n, j = 1, 2, ..., s], and then extract the required feature types;

[0039] Q24. Obtain the characteristic objective function;

[0040]

[0041] In the formula, m is the category dimension of the feature, and μ ik For the weighting of each feature, d ik denoted as Euclidean distance between feature nodes and feature centers, where c is the number of meaningful feature categories extracted from m categories, 1 < c < m;

[0042] (d ik ) 2 =||x k -V i || 2 , where X kThe feature node is the k-th item in the X matrix; v i The central node has an initial value that is a predefined empirical value, and subsequent values ​​are iteratively calculated from each iteration.

[0043] and

[0044] Q25. Construct the optimal objective function:

[0045] Q26. Obtain the globally optimal solution.

[0046] This invention also provides a multi-source detection fusion algorithm for a multi-layered interception and defense system, characterized by comprising the following steps:

[0047] W1. Detect moving targets using multi-source detectors to obtain different feature information;

[0048] Detection methods include Doppler radar detection, radar imaging detection, and infrared detection;

[0049] W2. Perform multi-objective information fusion processing, which includes the following steps:

[0050] W21, target discrimination algorithm;

[0051] U1 = [a1, a2, a3, a4, ... an];

[0052] U2 = [b1, b2, b3, b4, ... bn];

[0053] U3 = [c1,c2,c3,c4,...cn];

[0054] X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn]

[0055] U1, U2, and U3 are the information matrices obtained by the three detection methods, respectively.

[0056] X is a matrix containing all the information obtained from the three detection methods;

[0057] W22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n.

[0058] W23. Select data with similar characteristics for mapping and generalization processing to obtain the detection information function V=[Vij|i=1,2...,n,j=1,2,...,s], and then extract the required feature types;

[0059] W24. Obtain the characteristic objective function;

[0060]

[0061] In the formula, m is the feature category dimension, μik is the weight of each feature, represents the Euclidean distance between the feature node and the feature center, and c is the number of meaningful feature categories extracted from m categories, 1 < c < m;

[0062] (d ik ) 2 =||x k -V i || 2 , where Xk is the feature node, which is the kth item in the X matrix; vi is the center node, whose initial value is a specified empirical value, and the subsequent value is the value obtained from each calculation iteratively;

[0063] and

[0064] W25. Construct the optimal objective function:

[0065] W26. The globally optimal solution is obtained.

[0066] This invention also provides a multi-source detection fusion algorithm for a multi-layered interception and defense system, characterized by comprising the following steps:

[0067] R1. Detect moving targets using multi-source detectors to obtain different feature information;

[0068] Detection methods include Doppler radar detection, optical detection, and infrared detection;

[0069] R2. Perform multi-objective information fusion processing, which includes the following steps:

[0070] R21, target discrimination algorithm;

[0071] U1 = [a1, a2, a3, a4, ... an];

[0072] U2 = [b1, b2, b3, b4, ... bn];

[0073] U3 = [c1,c2,c3,c4,...cn];

[0074] X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn]

[0075] U1, U2, and U3 are the information matrices obtained by the three detection methods, respectively.

[0076] X is a matrix containing all the information obtained from the three detection methods;

[0077] R22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n.

[0078] R23. Select data with similar characteristics for mapping and generalization processing to obtain the detection information function V = [Vij|i = 1, 2, ..., n, j = 1, 2, ..., s], and then extract the required feature types;

[0079] R24. Obtain the characteristic objective function;

[0080]

[0081] In the formula, m is the category dimension of the feature, and μ ik The weighted sum for each feature represents the Euclidean distance between the feature node and the feature center, where c is the number of meaningful feature categories extracted from m categories, 1 < c < m;

[0082] (d ik ) 2 =||x k -V i || 2 , where Xk is the feature node, which is the kth item in the X matrix; vi is the center node, whose initial value is a specified empirical value, and the subsequent value is the value obtained from each calculation iteratively;

[0083] and

[0084] R25. Construct the optimal objective function:

[0085] R26, the globally optimal solution is obtained.

[0086] In the multi-source detection fusion algorithm of the multi-layered interception and defense system described above, the required feature types refer to the process of manually removing some detection data with low reference value based on the current specific conditions (environment, equipment status, etc.) and retaining valuable data to obtain the required detection information; and setting the data to be removed under a certain environment so that this setting can be applied in the same environment in the future.

[0087] Furthermore, the multi-source detection fusion algorithm of the above-mentioned multi-layered interception and defense system is characterized in that the optimal objective function is:

[0088]

[0089] Based on the experience of detection fusion and the verification of a large number of experiments and examples, the optimal objective function value is related to the different Euclidean distances of the same feature detection data. The Euclidean distance corresponds to the average error mean of each type of detection data. The closer the error mean is, the higher the weighting value represented by the objective function.

[0090] Furthermore, the globally optimal solution is:

[0091]

[0092] Compared with existing technologies, the multi-source detection fusion algorithm of the multi-layered interception and defense system provided by this invention can fuse detection information from various detectors, perform more diverse feature fusion selection, and get closer to the optimal detection value for the same features, thus making it easier to detect and identify targets.

[0093] The selected typical features, after being processed by the fusion algorithm of this invention, more closely approximate the true values. This invention can supplement the detection range functions lacking in Doppler radar detection; compared to optical detection, it can supplement detection data over a larger detection range, enabling earlier target identification; compared to radar imaging detection, it can supplement target angle information; and compared to infrared detection, it can supplement target angle information. Detailed Implementation

[0094] The technical solution adopted in this invention is a multi-source detection fusion algorithm for a multi-layered interception and defense system, including the following steps:

[0095] S1. Detect moving targets using multi-source detectors to obtain different feature information;

[0096] Detection methods include Doppler radar detection, radar imaging detection, optical detection, and infrared detection;

[0097] S2. Perform multi-objective information fusion processing, which includes the following steps:

[0098] S21, Target Differentiation Algorithm;

[0099] U1 = [a1, a2, a3, a4, ... an];

[0100] U2 = [b1, b2, b3, b4, ... bn];

[0101] U3 = [c1,c2,c3,c4,...cn];

[0102] U4 = [d1, d2, d3, d4, ... dn];

[0103] X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn,d1,d2,d3,d4,...dn]

[0104] U1, U2, U3, and U4 are the information matrices obtained from the four detection methods, respectively.

[0105] X is a matrix containing all the information obtained from the four detection methods;

[0106] S22. Divide the sampled data at each sampling time into categories according to their characteristics (arranged with time as the row and category as the column), and divide the detection data to be analyzed into m categories, 1 < m < n.

[0107] S23. Select data with similar characteristics and perform mapping generalization processing to obtain the detection information function V=[Vij|i=1,2...,n,j=1,2,...,s], and then extract the required feature types;

[0108] S24. Obtain the characteristic objective function;

[0109]

[0110] In the formula, m is the category dimension of the feature, and μ ik For the weighting of each feature, d ik denoted as Euclidean distance between feature nodes and feature centers, where c is the number of meaningful feature categories extracted from m categories, 1 < c < m;

[0111] (d ik ) 2 =||x k -V i || 2 , where X k The feature node is the k-th item in the X matrix; v i The central node is initialized with an assigned empirical value, and the optimal solution obtained from each subsequent calculation is used for iteration.

[0112] and

[0113] S25. Construct the optimal objective function:

[0114] S26. Obtain the global optimal solution.

[0115] This invention also provides a multi-source detection fusion algorithm for a multi-layered interception and defense system, characterized by comprising the following steps:

[0116] Q1. Detect moving targets using multi-source detectors to obtain different feature information;

[0117] Detection methods include Doppler radar detection, radar imaging detection, and optical detection;

[0118] Q2. Perform multi-objective information fusion processing, which includes the following steps:

[0119] Q21. Target discrimination algorithm;

[0120] U1 = [a1, a2, a3, a4, ... an];

[0121] U2 = [b1, b2, b3, b4, ... bn];

[0122] U3 = [c1,c2,c3,c4,...cn];

[0123] X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn]

[0124] U1, U2, and U3 are the information matrices obtained by the three detection methods, respectively.

[0125] X is a matrix containing all the information obtained from the three detection methods;

[0126] Q22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n.

[0127] Q23. Select data with similar characteristics and perform mapping generalization to obtain the detection information function V = [Vij|i = 1, 2, ..., n, j = 1, 2, ..., s], and then extract the required feature types;

[0128] Q24. Obtain the characteristic objective function;

[0129]

[0130] In the formula, m is the category dimension of the feature, and μ ik For the weighting of each feature, d ik denoted as Euclidean distance between feature nodes and feature centers, where c is the number of meaningful feature categories extracted from m categories, 1 < c < m;

[0131] (d ik ) 2 =||x k -V i || 2 , where X k The feature node is the k-th item in the X matrix; v i The central node has an initial value that is a predefined empirical value, and subsequent values ​​are iteratively calculated from each iteration.

[0132] and

[0133] Q25. Construct the optimal objective function:

[0134] Q26. Obtain the globally optimal solution.

[0135] This invention also provides a multi-source detection fusion algorithm for a multi-layered interception and defense system, characterized by comprising the following steps:

[0136] W1. Detect moving targets using multi-source detectors to obtain different feature information;

[0137] Detection methods include Doppler radar detection, radar imaging detection, and infrared detection;

[0138] W2. Perform multi-objective information fusion processing, which includes the following steps:

[0139] W21, target discrimination algorithm;

[0140] U1 = [a1, a2, a3, a4, ... an];

[0141] U2 = [b1, b2, b3, b4, ... bn];

[0142] U3 = [c1,c2,c3,c4,...cn];

[0143] X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn]

[0144] U1, U2, and U3 are the information matrices obtained by the three detection methods, respectively.

[0145] X is a matrix containing all the information obtained from the three detection methods;

[0146] W22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n.

[0147] W23. Select data with similar characteristics for mapping and generalization processing to obtain the detection information function V=[Vij|i=1,2...,n,j=1,2,...,s], and then extract the required feature types;

[0148] W24. Obtain the characteristic objective function;

[0149]

[0150] In the formula, m is the feature category dimension, μik is the weight of each feature, represents the Euclidean distance between the feature node and the feature center, and c is the number of meaningful feature categories extracted from m categories, 1 < c < m;

[0151] (d ik ) 2 =||x k -Vi || 2 , where Xk is the feature node, which is the kth item in the X matrix; vi is the center node, whose initial value is a specified empirical value, and the subsequent value is the value obtained from each calculation iteratively;

[0152] and

[0153] W25. Construct the optimal objective function:

[0154] W26. The globally optimal solution is obtained.

[0155] This invention also provides a multi-source detection fusion algorithm for a multi-layered interception and defense system, characterized by comprising the following steps:

[0156] R1. Detect moving targets using multi-source detectors to obtain different feature information;

[0157] Detection methods include Doppler radar detection, optical detection, and infrared detection;

[0158] R2. Perform multi-objective information fusion processing, which includes the following steps:

[0159] R21, target discrimination algorithm;

[0160] U1 = [a1, a2, a3, a4, ... an];

[0161] U2 = [b1, b2, b3, b4, ... bn];

[0162] U3 = [c1,c2,c3,c4,...cn];

[0163] X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn]

[0164] U1, U2, and U3 are the information matrices obtained by the three detection methods, respectively.

[0165] X is a matrix containing all the information obtained from the three detection methods;

[0166] R22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n.

[0167] R23. Select data with similar characteristics for mapping and generalization processing to obtain the detection information function V = [Vij|i = 1, 2, ..., n, j = 1, 2, ..., s], and then extract the required feature types;

[0168] R24. Obtain the characteristic objective function;

[0169]

[0170] In the formula, m is the category dimension of the feature, and μ ik The weighted sum for each feature represents the Euclidean distance between the feature node and the feature center, where c is the number of meaningful feature categories extracted from m categories, 1 < c < m;

[0171] (d ik ) 2 =||x k -V i || 2 , where Xk is the feature node, which is the kth item in the X matrix; vi is the center node, whose initial value is a specified empirical value, and the subsequent value is the value obtained from each calculation iteratively;

[0172] and

[0173] R25. Construct the optimal objective function:

[0174] R26, the globally optimal solution is obtained.

[0175] In the above steps, the required feature types refer to the process of manually removing some detection data with low reference value based on the current specific conditions (environment, equipment status, etc.) and retaining valuable data to obtain the required detection information; and setting the data to be removed under a certain environment so that this setting can be applied in the same environment in the future.

[0176] Furthermore, in the multi-source detection fusion algorithm of the aforementioned multi-layered interception and defense system, the optimal objective function is:

[0177]

[0178] Based on the experience of detection fusion and the verification of a large number of experiments and examples, the optimal objective function value is related to the different Euclidean distances of the same feature detection data. The Euclidean distance corresponds to the average error mean of each type of detection data. The closer the error mean is, the higher the weighting value represented by the objective function.

[0179] Furthermore, the feature center of the global optimal solution is:

[0180]

[0181] The typical features selected by the algorithm of this invention, after data processing, are closer to the true value, which can supplement the detection range function not found by Doppler radar detection, and the target angle information can be more reliable to the radar detection value. That is, the data obtained by Doppler radar detection in this type of feature has a higher weight.

[0182] Compared to optical detection, it can supplement detection data over a wider detection range, enabling earlier target identification. In terms of imaging accuracy, it places greater trust in optical detection values, meaning that data obtained through optical detection has higher weighting in such features.

[0183] Compared to radar imaging detection, it can supplement target angle information. For the size of long-range targets, radar imaging detection values ​​are more reliable; that is, data obtained from radar imaging detection has higher weighting in such features.

[0184] Compared to infrared detection, it can supplement target angle information. In terms of temperature characteristics, infrared detection is more reliable, meaning that data obtained from infrared detection has higher weighting in this type of feature.

[0185] After calculating the optimal solution (optimal target information) through the fusion algorithm, the strike strategy mainly adopts the conditional threshold algorithm: mainly based on distance as the threshold, where S is the distance between the target and the system;

[0186] Upon receiving enemy intelligence:

[0187] When 10km≤S≤15km, the electronic jamming equipment on the ground electromagnetic countermeasures vehicle starts to work.

[0188] When S≤10km, the laser weapon begins to irradiate and damage the target;

[0189] When the distance between the microwave-damaged UAV and the target is S≤3km, microwave damage is performed on the target.

[0190] When S≤2km, the kinetic energy interceptor intercepts and destroys the target;

[0191] When S≤200m, release the smoke grenade, and the radar and infrared decoy will start working.

[0192] The multi-source detection fusion algorithm for a multi-layered interception and defense system provided by this invention is not limited to three or four detection methods. More types of detection methods can also use the algorithm of this invention to perform multi-source information fusion.

Claims

1. A multi-source detection fusion method for a multi-layered interception and defense system, characterized in that, Includes the following steps: S1. Detect moving targets using multi-source detectors to obtain different feature information; Detection methods include Doppler radar detection, radar imaging detection, optical detection, and infrared detection; S2. Perform multi-objective information fusion processing, which includes the following steps: S21, Target Differentiation Algorithm; U1 = [a1, a2, a3, a4, ... an]; U2 = [b1, b2, b3, b4, ... bn]; U3 = [c1,c2,c3,c4,...cn]; U4 = [d1, d2, d3, d4, ... dn]; X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn,d1,d2,d3,d4,...dn] U1, U2, U3, and U4 are the information matrices obtained from the four detection methods, respectively. X is a matrix containing all the information obtained from the four detection methods; S22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n. S23. Select data with similar characteristics and perform mapping generalization processing to obtain the detection information function V=[Vij|i=1,2...,n,j=1,2,...,s], and then extract the required feature types; S24. Obtain the characteristic objective function; In the formula, m is the category dimension of the feature, and μ ik For the weighting of each feature, d ik represents the Euclidean distance between the feature node and the feature center, where 1 < c < m; (d ik ) 2 =||x k -V i || 2 , where x k The feature node is the k-th item in the X matrix; v i The central node is initialized with an assigned empirical value, and subsequent iterations use the optimal solution obtained from each calculation; and S25. The optimal objective function is constructed as follows: S26. Obtain the global optimal solution.

2. A multi-source detection fusion method for a multi-layered interception and defense system, characterized in that, Includes the following steps: Q1. Detect moving targets using multi-source detectors to obtain different feature information; Detection methods include Doppler radar detection, radar imaging detection, and optical detection; Q2. Perform multi-objective information fusion processing, which includes the following steps: Q21. Target discrimination algorithm; U1 = [a1, a2, a3, a4, ... an]; U2 = [b1, b2, b3, b4, ... bn]; U3 = [c1,c2,c3,c4,...cn]; X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn] U1, U2, and U3 are the information matrices obtained by the three detection methods, respectively. X is a matrix containing all the information obtained from the three detection methods; Q22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n. Q23. Select data with similar characteristics and perform mapping generalization to obtain the detection information function V = [Vij|i = 1, 2, ..., n, j = 1, 2, ..., s], and then extract the required feature types; Q24. Obtain the characteristic objective function; In the formula, m is the category dimension of the feature, and μ ik For the weighting of each feature, d ik represents the Euclidean distance between the feature node and the feature center, where 1 < c < m; (d ik ) 2 =||x k -V i || 2 , where x k The feature node is the k-th item in the X matrix; v i The central node has an initial value that is a predefined empirical value, and subsequent values ​​are iteratively calculated from each iteration. and Q25. Construct the optimal objective function as follows: Q26. Obtain the globally optimal solution.

3. A multi-source detection fusion method for a multi-layered interception and defense system, characterized in that, Includes the following steps: W1. Detect moving targets using multi-source detectors to obtain different feature information; Detection methods include Doppler radar detection, radar imaging detection, and infrared detection; W2. Perform multi-objective information fusion processing, which includes the following steps: W21, target discrimination algorithm; U1 = [a1, a2, a3, a4, ... an]; U2 = [b1, b2, b3, b4, ... bn]; U3 = [c1,c2,c3,c4,...cn]; X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn] U1, U2, and U3 are the information matrices obtained by the three detection methods, respectively. X is a matrix containing all the information obtained from the three detection methods; W22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n. W23. Select data with similar characteristics for mapping and generalization processing to obtain the detection information function V=[Vij|i=1,2...,n,j=1,2,...,s], and then extract the required feature types; W24. Obtain the characteristic objective function; In the formula, m is the feature category dimension, μik is the weight of each feature, and represents the Euclidean distance between the feature node and the feature center, where 1 < c < m; (d ik ) 2 =||x k -V i || 2 , where x k is the feature node, which is the k-th item in the X matrix; vi is the center node, whose initial value is a specified empirical value, and the subsequent values ​​are iterative values ​​obtained from each calculation; and W25. The optimal objective function is constructed as follows: W26. The globally optimal solution is obtained.

4. A multi-source detection fusion method for a multi-layered interception and defense system, characterized in that, Includes the following steps: R1. Detect moving targets using multi-source detectors to obtain different feature information; Detection methods include Doppler radar detection, optical detection, and infrared detection; R2. Perform multi-objective information fusion processing, which includes the following steps: R21, target discrimination algorithm; U1 = [a1, a2, a3, a4, ... an]; U2 = [b1, b2, b3, b4, ... bn]; U3 = [c1,c2,c3,c4,...cn]; X=[a1,a2,a3,a4,...an,b1,b2,b3,b4,...bn,c1,c2,c3,c4,...cn] U1, U2, and U3 are the information matrices obtained by the three detection methods, respectively. X is a matrix containing all the information obtained from the three detection methods; R22. Divide the sampled data at each sampling time into types according to their characteristics, and divide the detection data to be analyzed into m types, where 1 < m < n. R23. Select data with similar characteristics for mapping and generalization processing to obtain the detection information function V = [Vij|i = 1, 2, ..., n, j = 1, 2, ..., s], and then extract the required feature types; R24. Obtain the characteristic objective function; In the formula, m is the category dimension of the feature, and μ ik For each feature, the weighted sum represents the Euclidean distance between the feature node and the feature center, where 1 < c < m; (d ik ) 2 =||x k -V i || 2 , where x k The feature node is the k-th item in the X matrix; v i The central node has an initial value that is a predefined empirical value, and subsequent values ​​are iteratively calculated from each iteration. and R25. The optimal objective function is constructed as follows: R26, the globally optimal solution is obtained.

5. A multi-source detection fusion method for a multi-layered interception and defense system according to any one of claims 1-4, characterized in that, The globally optimal solution is:

Images